diff --git "a/data/it/test.jsonl" "b/data/it/test.jsonl" new file mode 100644--- /dev/null +++ "b/data/it/test.jsonl" @@ -0,0 +1,618 @@ +{"source": "Incremental class learning involves sequentially learning classes in bursts of examples from the same class. This violates the assumptions that underlie methods for training standard deep neural networks, and will cause them to suffer from catastrophic forgetting. Arguably, the best method for incremental class learning is iCaRL, but it requires storing training examples for each class, making it challenging to scale. Here, we propose FearNet for incremental class learning. FearNet is a generative model that does not store previous examples, making it memory efficient. FearNet uses a brain-inspired dual-memory system in which new memories are consolidated from a network for recent memories inspired by the mammalian hippocampal complex to a network for long-term storage inspired by medial prefrontal cortex. Memory consolidation is inspired by mechanisms that occur during sleep. FearNet also uses a module inspired by the basolateral amygdala for determining which memory system to use for recall. FearNet achieves state-of-the-art performance at incremental class learning on image (CIFAR-100, CUB-200) and audio classification (AudioSet) benchmarks.", "target": ["FearNet è una rete neurale efficiente in termini di memoria, ispirata alla formazione della memoria nel cervello dei mammiferi, che è in grado di apprendere classi incrementali senza catastrophic forgetting.", "Questo articolo presenta una nuova soluzione a un problema di classificazione incrementale basato su un sistema a doppia memoria."]} +{"source": "Multi-view learning can provide self-supervision when different views are available of the same data. Distributional hypothesis provides another form of useful self-supervision from adjacent sentences which are plentiful in large unlabelled corpora. Motivated by the asymmetry in the two hemispheres of the human brain as well as the observation that different learning architectures tend to emphasise different aspects of sentence meaning, we present two multi-view frameworks for learning sentence representations in an unsupervised fashion. One framework uses a generative objective and the other a discriminative one. In both frameworks, the final representation is an ensemble of two views, in which, one view encodes the input sentence with a Recurrent Neural Network (RNN), and the other view encodes it with a simple linear model. We show that, after learning, the vectors produced by our multi-view frameworks provide improved representations over their single-view learnt counterparts, and the combination of different views gives representational improvement over each view and demonstrates solid transferability on standard downstream tasks.", "target": ["L'apprendimento multi-view migliora l'apprendimento unsupervised di sentence embedding", "L'approccio utilizza diversi encoder complementari della frase in input e la massimizzazione del consenso.", "L'articolo presenta un quadro multi-view per migliorare la rappresentazione delle frasi nei task NLP usando architetture generative e discriminative.", "Questo articolo mostra che le strutture multi-view sono più efficaci dell'uso di encoder individuali per l'apprendimento di rappresentazioni di frasi."]} +{"source": "We show how discrete objects can be learnt in an unsupervised fashion from pixels, and how to perform reinforcement learning using this object representation. More precisely, we construct a differentiable mapping from an image to a discrete tabular list of objects, where each object consists of a differentiable position, feature vector, and scalar presence value that allows the representation to be learnt using an attention mechanism. Applying this mapping to Atari games, together with an interaction net-style architecture for calculating quantities from objects, we construct agents that can play Atari games using objects learnt in an unsupervised fashion. During training, many natural objects emerge, such as the ball and paddles in Pong, and the submarine and fish in Seaquest. This gives the first reinforcement learning agent for Atari with an interpretable object representation, and opens the avenue for agents that can conduct object-based exploration and generalization.", "target": ["Mostriamo come gli oggetti discreti possono essere appresi in modo unsupervised dai pixel, e come eseguire reinforcement learning usando questa rappresentazione degli oggetti.", "Un metodo per imparare rappresentazioni di oggetti dai pixel per fare reinforcement learning.", "L'articolo propone un'architettura neurale per mappare i flussi video su una collezione discreta di oggetti, senza annotazioni umane, utilizzando una loss di ricostruzione dei pixel non supervisionata."]} +{"source": "Most recent gains in visual recognition have originated from the inclusion of attention mechanisms in deep convolutional networks (DCNs). Because these networks are optimized for object recognition, they learn where to attend using only a weak form of supervision derived from image class labels. Here, we demonstrate the benefit of using stronger supervisory signals by teaching DCNs to attend to image regions that humans deem important for object recognition. We first describe a large-scale online experiment (ClickMe) used to supplement ImageNet with nearly half a million human-derived \"top-down\" attention maps. Using human psychophysics, we confirm that the identified top-down features from ClickMe are more diagnostic than \"bottom-up\" saliency features for rapid image categorization. As a proof of concept, we extend a state-of-the-art attention network and demonstrate that adding ClickMe supervision significantly improves its accuracy and yields visual features that are more interpretable and more similar to those used by human observers.", "target": ["Un dataset su larga scala per training di modelli di attention per il riconoscimento degli oggetti porta a un riconoscimento degli oggetti più accurato, interpretabile e simile a quello umano.", "Il paper sostiene che i recenti guadagni nel riconoscimento visivo derivino dall'utilizzo di meccanismi di attention visiva nelle deep convolutional network, che imparano dove concentrarsi attraverso una forma debole di supervisione basata sulle label di classe delle immagini.", "Presenta un nuovo approccio all'attention in cui un grande dataset sull'attention viene raccolto e utilizzato per addestrare una NN in modo supervisionato per sfruttare l'attention umana self-reported.", "Questo articolo propone un nuovo approccio per utilizzare segnali più informativi, in particolare, le regioni che gli esseri umani ritengono importanti nelle immagini, per migliorare le deep convolutional neural network."]} +{"source": "In recent years, deep neural networks have demonstrated outstanding performancein many machine learning tasks. However, researchers have discovered that thesestate-of-the-art models are vulnerable to adversarial examples: legitimate examples added by small perturbations which are unnoticeable to human eyes. Adversarial training, which augments the training data with adversarial examples duringthe training process, is a well known defense to improve the robustness of themodel against adversarial attacks. However, this robustness is only effective tothe same attack method used for adversarial training. Madry et al. (2017) suggest that effectiveness of iterative multi-step adversarial attacks and particularlythat projected gradient descent (PGD) may be considered the universal first order adversary and applying the adversarial training with PGD implies resistanceagainst many other first order attacks. However, the computational cost of theadversarial training with PGD and other multi-step adversarial examples is muchhigher than that of the adversarial training with other simpler attack techniques. In this paper, we show how strong adversarial examples can be generated only ata cost similar to that of two runs of the fast gradient sign method (FGSM), allowing defense against adversarial attacks with a robustness level comparable to thatof the adversarial training with multi-step adversarial examples. We empiricallydemonstrate the effectiveness of the proposed two-step defense approach againstdifferent attack methods and its improvements over existing defense strategies.", "target": ["Abbiamo proposto un metodo di difesa efficiente in termini di tempo contro gli adversarial attack one-step e iterativi.", "Proporre un nuovo metodo efficiente dal punto di vista computazionale chiamato e2SAD che genera insiemi di due sample di adversarial training per ogni sample di training pulito.", "L'articolo introduce un metodo di adversarial defense in due fasi, per generare due adversarial example per ogni sample pulito e includerli nel ciclo di training attuale per raggiungere la robustezza e affermare che può superare i metodi iterativi più costosi.", "L'articolo presenta un approccio in 2 fasi per generare forti adversarial example ad un costo molto inferiore rispetto ai recenti adversarial attack iterativi in più fasi."]} +{"source": "Recently several different deep learning architectures have been proposed that take a string of characters as the raw input signal and automatically derive features for text classification. Little studies are available that compare the effectiveness of these approaches for character based text classification with each other. In this paper we perform such an empirical comparison for the important cybersecurity problem of DGA detection: classifying domain names as either benign vs. produced by malware (i.e., by a Domain Generation Algorithm). Training and evaluating on a dataset with 2M domain names shows that there is surprisingly little difference between various convolutional neural network (CNN) and recurrent neural network (RNN) based architectures in terms of accuracy, prompting a preference for the simpler architectures, since they are faster to train and less prone to overfitting.", "target": ["Un confronto tra cinque architetture di deep neural network per il rilevamento di domain name malevoli mostra sorprendentemente poche differenze.", "Gli autori propongono di utilizzare cinque architetture di deep learning per il task di cybersicurezza relativo al rilevamento degli algoritmi di generazione del dominio.", "Applica diverse architetture NN per classificare gli URL tra quelli benigni e quelli relativi a malware.", "Questo articolo propone di riconoscere automaticamente i nomi di dominio come malevoli o benigni tramite deep network addestrate per classificare direttamente la sequenza di caratteri."]} +{"source": "Recognizing the relationship between two texts is an important aspect of natural language understanding (NLU), and a variety of neural network models have been proposed for solving NLU tasks. Unfortunately, recent work showed that the datasets these models are trained on often contain biases that allow models to achieve non-trivial performance without possibly learning the relationship between the two texts. We propose a framework for building robust models by using adversarial learning to encourage models to learn latent, bias-free representations. We test our approach in a Natural Language Inference (NLI) scenario, and show that our adversarially-trained models learn robust representations that ignore known dataset-specific biases. Our experiments demonstrate that our models are more robust to new NLI datasets.", "target": ["I metodi di adversarial training incoraggiano i modelli di NLI a ignorare i bias specifici del dataset e aiutano i modelli a fare transfer attraverso i dataset.", "L'articolo propone un adversarial setup per mitigare gli artefatti di annotazione nei dati di natural language inference", "Questo articolo presenta un metodo per rimuovere i bias di un modello di entailment testuale attraverso un obiettivo di adversarial training."]} +{"source": "We study the problem of learning representations of entities and relations in knowledge graphs for predicting missing links. The success of such a task heavily relies on the ability of modeling and inferring the patterns of (or between) the relations. In this paper, we present a new approach for knowledge graph embedding called RotatE, which is able to model and infer various relation patterns including: symmetry/antisymmetry, inversion, and composition. Specifically, the RotatE model defines each relation as a rotation from the source entity to the target entity in the complex vector space. In addition, we propose a novel self-adversarial negative sampling technique for efficiently and effectively training the RotatE model. Experimental results on multiple benchmark knowledge graphs show that the proposed RotatE model is not only scalable, but also able to infer and model various relation patterns and significantly outperform existing state-of-the-art models for link prediction.", "target": ["Un nuovo approccio allo stato dell'arte per l'embedding dei knowledge graph.", "Presenta una funzione di scoring per neural link prediction che può dedurre modelli di simmetria, antisimmetria, inversione e composizione delle relazioni in una knowledge base.", "Questo articolo propone un approccio all'embedding dei knowledge graph modellando le relazioni come rotazioni nello spazio vettoriale complesso.", "Propone un metodo per l'embedding di grafi da utilizzare per la link prediction"]} +{"source": "Deep learning algorithms have been known to be vulnerable to adversarial perturbations in various tasks such as image classification. This problem was addressed by employing several defense methods for detection and rejection of particular types of attacks. However, training and manipulating networks according to particular defense schemes increases computational complexity of the learning algorithms. In this work, we propose a simple yet effective method to improve robustness of convolutional neural networks (CNNs) to adversarial attacks by using data dependent adaptive convolution kernels. To this end, we propose a new type of HyperNetwork in order to employ statistical properties of input data and features for computation of statistical adaptive maps. Then, we filter convolution weights of CNNs with the learned statistical maps to compute dynamic kernels. Thereby, weights and kernels are collectively optimized for learning of image classification models robust to adversarial attacks without employment of additional target detection and rejection algorithms. We empirically demonstrate that the proposed method enables CNNs to spontaneously defend against different types of attacks, e.g. attacks generated by Gaussian noise, fast gradient sign methods (Goodfellow et al., 2014) and a black-box attack (Narodytska & Kasiviswanathan, 2016).", "target": ["Abbiamo modificato la CNN usando HyperNetworks e abbiamo osservato una migliore robustezza contro gli adversarial example.", "Migliorare la robustezza e l'affidabilità delle deep convolutional neural network usando kernel di convoluzione dipendenti dai dati"]} +{"source": "Adapting deep networks to new concepts from a few examples is challenging, due to the high computational requirements of standard fine-tuning procedures. Most work on few-shot learning has thus focused on simple learning techniques for adaptation, such as nearest neighbours or gradient descent. Nonetheless, the machine learning literature contains a wealth of methods that learn non-deep models very efficiently. In this paper, we propose to use these fast convergent methods as the main adaptation mechanism for few-shot learning. The main idea is to teach a deep network to use standard machine learning tools, such as ridge regression, as part of its own internal model, enabling it to quickly adapt to novel data. This requires back-propagating errors through the solver steps. While normally the cost of the matrix operations involved in such a process would be significant, by using the Woodbury identity we can make the small number of examples work to our advantage. We propose both closed-form and iterative solvers, based on ridge regression and logistic regression components. Our methods constitute a simple and novel approach to the problem of few-shot learning and achieve performance competitive with or superior to the state of the art on three benchmarks.", "target": ["Proponiamo un approccio di meta-learning per la classificazione few shot che raggiunge forti prestazioni ad alta velocità facendo la back-propagation attraverso la soluzione di risolutori veloci, come la regressione ridge o la regressione logistica.", "L'articolo propone un algoritmo per il meta-learning che equivale a fissare le feature (cioè tutti gli hidden layer di una deep NN), e trattare ogni task come se avesse il proprio layer finale come una regressione ridge o una regressione logistica.", "Questo articolo propone un approccio di meta-learning per il problema della classificazione few shot, usano un metodo basato sulla parametrizzazione del learner per ogni task con un risolutore in forma chiusa."]} +{"source": "While many active learning papers assume that the learner can simply ask for a label and receive it, real annotation often presents a mismatch between the form of a label (say, one among many classes), and the form of an annotation (typically yes/no binary feedback). To annotate examples corpora for multiclass classification, we might need to ask multiple yes/no questions, exploiting a label hierarchy if one is available. To address this more realistic setting, we propose active learning with partial feedback (ALPF), where the learner must actively choose both which example to label and which binary question to ask. At each step, the learner selects an example, asking if it belongs to a chosen (possibly composite) class. Each answer eliminates some classes, leaving the learner with a partial label. The learner may then either ask more questions about the same example (until an exact label is uncovered) or move on immediately, leaving the first example partially labeled. Active learning with partial labels requires (i) a sampling strategy to choose (example, class) pairs, and (ii) learning from partial labels between rounds. Experiments on Tiny ImageNet demonstrate that our most effective method improves 26% (relative) in top-1 classification accuracy compared to i.i.d. baselines and standard active learners given 30% of the annotation budget that would be required (naively) to annotate the dataset. Moreover, ALPF-learners fully annotate TinyImageNet at 42% lower cost. Surprisingly, we observe that accounting for per-example annotation costs can alter the conventional wisdom that active learners should solicit labels for hard examples.", "target": ["Forniamo una nuova prospettiva sul training di un modello di apprendimento automatico da zero nel setting delle label gerarchiche, cioè pensandolo come una comunicazione a due vie tra l'uomo e gli algoritmi, e studiamo come possiamo sia misurare che migliorare l'efficienza.", "Introduce un nuovo setting di Active Learning in cui l'oracolo offre una label parziale o debole invece di cercare la label di un particolare esempio, portando a un recupero più semplice delle informazioni.", "Questo articolo propone un metodo di active learning con feedback parziale che supera le baseline esistenti sotto un budget limitato.", "L'articolo considera un problema di classificazione multiclasse in cui le label sono raggruppate in un dato numero M di sottoinsiemi, che contengono tutte le label individuali come singleton."]} +{"source": "Despite their prevalence, Euclidean embeddings of data are fundamentally limited in their ability to capture latent semantic structures, which need not conform to Euclidean spatial assumptions. Here we consider an alternative, which embeds data as discrete probability distributions in a Wasserstein space, endowed with an optimal transport metric. Wasserstein spaces are much larger and more flexible than Euclidean spaces, in that they can successfully embed a wider variety of metric structures. We propose to exploit this flexibility by learning an embedding that captures the semantic information in the Wasserstein distance between embedded distributions. We examine empirically the representational capacity of such learned Wasserstein embeddings, showing that they can embed a wide variety of complex metric structures with smaller distortion than an equivalent Euclidean embedding. We also investigate an application to word embedding, demonstrating a unique advantage of Wasserstein embeddings: we can directly visualize the high-dimensional embedding, as it is a probability distribution on a low-dimensional space. This obviates the need for dimensionality reduction techniques such as t-SNE for visualization.", "target": ["Mostriamo che gli spazi di Wasserstein sono buoni obiettivi per incorporare dati con una struttura semantica complessa.", "Impara gli embedding in uno spazio discreto di distribuzioni di probabilità, utilizzando una versione minimizzata e regolarizzata delle distanze di Wasserstein.", "L'articolo descrive un nuovo metodo di embedding che incorpora i dati nello spazio delle misure di probabilità dotato della distanza di Wasserstein.", "L'articolo propone di incorporare i dati in spazi Wasserstein a bassa dimensione, che possono catturare la struttura sottostante ai dati in modo più accurato."]} +{"source": "Clustering high-dimensional datasets is hard because interpoint distances become less informative in high-dimensional spaces. We present a clustering algorithm that performs nonlinear dimensionality reduction and clustering jointly. The data is embedded into a lower-dimensional space by a deep autoencoder. The autoencoder is optimized as part of the clustering process. The resulting network produces clustered data. The presented approach does not rely on prior knowledge of the number of ground-truth clusters. Joint nonlinear dimensionality reduction and clustering are formulated as optimization of a global continuous objective. We thus avoid discrete reconfigurations of the objective that characterize prior clustering algorithms. Experiments on datasets from multiple domains demonstrate that the presented algorithm outperforms state-of-the-art clustering schemes, including recent methods that use deep networks.", "target": ["Un algoritmo di clustering che esegue la riduzione non lineare della dimensionalità e il clustering congiunto ottimizzando un obiettivo globale continuo.", "Presenta un algoritmo di clustering risolvendo congiuntamente l'autoencoder profondo e il clustering come obiettivo continuo globale, mostrando risultati migliori rispetto agli schemi di clustering allo stato dell'arte.", "Deep Continuous Clustering è un metodo di clustering che integra l'obiettivo dell'autoencoder con l'obiettivo del clustering e poi si addestra usando SGD."]} +{"source": "Deep convolutional neural networks (CNNs) are deployed in various applications but demand immense computational requirements. Pruning techniques and Winograd convolution are two typical methods to reduce the CNN computation. However, they cannot be directly combined because Winograd transformation fills in the sparsity resulting from pruning. Li et al. (2017) propose sparse Winograd convolution in which weights are directly pruned in the Winograd domain, but this technique is not very practical because Winograd-domain retraining requires low learning rates and hence significantly longer training time. Besides, Liu et al. (2018) move the ReLU function into the Winograd domain, which can help increase the weight sparsity but requires changes in the network structure. To achieve a high Winograd-domain weight sparsity without changing network structures, we propose a new pruning method, spatial-Winograd pruning. As the first step, spatial-domain weights are pruned in a structured way, which efficiently transfers the spatial-domain sparsity into the Winograd domain and avoids Winograd-domain retraining. For the next step, we also perform pruning and retraining directly in the Winograd domain but propose to use an importance factor matrix to adjust weight importance and weight gradients. This adjustment makes it possible to effectively retrain the pruned Winograd-domain network without changing the network structure. For the three models on the datasets of CIFAR-10, CIFAR-100, and ImageNet, our proposed method can achieve the Winograd-domain sparsities of 63%, 50%, and 74%, respectively.", "target": ["Per accelerare il calcolo delle reti neurali convoluzionali, proponiamo una nuova tecnica di pruning in due fasi che raggiunge una maggiore sparsità dei pesi nel dominio di Winograd senza cambiare la struttura della rete.", "Propone un quadro di pruning spatial-Winograd che permette di mantenere il peso pruned dal dominio spaziale nel dominio di Winograd e migliora la sparsità del dominio di Winograd.", "Propone due tecniche per il pruning dei layer convoluzionali che utilizzano l'algoritmo di Winograd"]} +{"source": "In federated learning problems, data is scattered across different servers and exchanging or pooling it is often impractical or prohibited. We develop a Bayesian nonparametric framework for federated learning with neural networks. Each data server is assumed to train local neural network weights, which are modeled through our framework. We then develop an inference approach that allows us to synthesize a more expressive global network without additional supervision or data pooling. We then demonstrate the efficacy of our approach on federated learning problems simulated from two popular image classification datasets.", "target": ["Proponiamo un modello bayesiano non parametrico per federated learning con reti neurali.", "Utilizza il processo beta per fare neural federated matching.", "L'articolo considera il federated learning delle reti neurali, dove i dati sono distribuiti su più macchine e l'allocazione dei data point è potenzialmente disomogenea e sbilanciata."]} +{"source": "We present a general-purpose method to train Markov chain Monte Carlo kernels, parameterized by deep neural networks, that converge and mix quickly to their target distribution. Our method generalizes Hamiltonian Monte Carlo and is trained to maximize expected squared jumped distance, a proxy for mixing speed. We demonstrate large empirical gains on a collection of simple but challenging distributions, for instance achieving a 106x improvement in effective sample size in one case, and mixing when standard HMC makes no measurable progress in a second. Finally, we show quantitative and qualitative gains on a real-world task: latent-variable generative modeling. Python source code will be open-sourced with the camera-ready paper.", "target": ["Metodo generale per addestrare kernel MCMC espressivi parametrizzati con deep neural network. Data una distribuzione di destinazione p, il nostro metodo fornisce un sampler veloce, in grado di esplorare in modo efficiente lo spazio di stato.", "Propone un HMC generalizzato modificando l'integratore leapfrog utilizzando reti neurali per far convergere e mixare rapidamente il sampler."]} +{"source": "This paper addresses the problem of evaluating learning systems in safety critical domains such as autonomous driving, where failures can have catastrophic consequences. We focus on two problems: searching for scenarios when learned agents fail and assessing their probability of failure. The standard method for agent evaluation in reinforcement learning, Vanilla Monte Carlo, can miss failures entirely, leading to the deployment of unsafe agents. We demonstrate this is an issue for current agents, where even matching the compute used for training is sometimes insufficient for evaluation. To address this shortcoming, we draw upon the rare event probability estimation literature and propose an adversarial evaluation approach. Our approach focuses evaluation on adversarially chosen situations, while still providing unbiased estimates of failure probabilities. The key difficulty is in identifying these adversarial situations -- since failures are rare there is little signal to drive optimization. To solve this we propose a continuation approach that learns failure modes in related but less robust agents. Our approach also allows reuse of data already collected for training the agent. We demonstrate the efficacy of adversarial evaluation on two standard domains: humanoid control and simulated driving. Experimental results show that our methods can find catastrophic failures and estimate failures rates of agents multiple orders of magnitude faster than standard evaluation schemes, in minutes to hours rather than days.", "target": ["Mostriamo che le failure rare ma catastrofiche possono essere interamente non rilevate dai test casuali, il che pone problemi per un deployment sicuro. Il nostro approccio proposto per gli adversarial test risolve questo problema.", "Propone un metodo che impara un predittore di probabilità di fallimento per un learned agent, portando a predire quali stati iniziali causano il fallimento di un sistema.", "Questo articolo propone un approccio di importance sampling per campionare i casi di fallimento per gli algoritmi RL basati su una funzione appresa tramite una rete neurale sui fallimenti che si verificano durante il training dell'agente", "Questo articolo ha proposto un approccio adversarial per identificare i casi di fallimento catastrofico nel reinforcement learning."]} +{"source": "The variational autoencoder (VAE) is a popular combination of deep latent variable model and accompanying variational learning technique. By using a neural inference network to approximate the model's posterior on latent variables, VAEs efficiently parameterize a lower bound on marginal data likelihood that can be optimized directly via gradient methods. In practice, however, VAE training often results in a degenerate local optimum known as \"posterior collapse\" where the model learns to ignore the latent variable and the approximate posterior mimics the prior. In this paper, we investigate posterior collapse from the perspective of training dynamics. We find that during the initial stages of training the inference network fails to approximate the model's true posterior, which is a moving target. As a result, the model is encouraged to ignore the latent encoding and posterior collapse occurs. Based on this observation, we propose an extremely simple modification to VAE training to reduce inference lag: depending on the model's current mutual information between latent variable and observation, we aggressively optimize the inference network before performing each model update. Despite introducing neither new model components nor significant complexity over basic VAE, our approach is able to avoid the problem of collapse that has plagued a large amount of previous work. Empirically, our approach outperforms strong autoregressive baselines on text and image benchmarks in terms of held-out likelihood, and is competitive with more complex techniques for avoiding collapse while being substantially faster.", "target": ["Per affrontare il posterior collapse nei VAE, proponiamo una procedura di training nuova ma semplice che ottimizza in modo aggressivo la rete di inferenza con più aggiornamenti. Questa nuova procedura di training attenua il posterior collapse e porta a un modello VAE migliore.", "Esamina il fenomeno del posterior collapse, mostrando che un maggiore training della rete di inferenza può ridurre il problema e portare a ottimi migliori.", "Gli autori propongono di cambiare la procedura di training delle VAE solo come soluzione al posterior collapse, lasciando il modello e l'obiettivo intatti."]} +{"source": "Online healthcare services can provide the general public with ubiquitous access to medical knowledge and reduce the information access cost for both individuals and societies. To promote these benefits, it is desired to effectively expand the scale of high-quality yet novel relational medical entity pairs that embody rich medical knowledge in a structured form. To fulfill this goal, we introduce a generative model called Conditional Relationship Variational Autoencoder (CRVAE), which can discover meaningful and novel relational medical entity pairs without the requirement of additional external knowledge. Rather than discriminatively identifying the relationship between two given medical entities in a free-text corpus, we directly model and understand medical relationships from diversely expressed medical entity pairs. The proposed model introduces the generative modeling capacity of variational autoencoder to entity pairs, and has the ability to discover new relational medical entity pairs solely based on the existing entity pairs. Beside entity pairs, relationship-enhanced entity representations are obtained as another appealing benefit of the proposed method. Both quantitative and qualitative evaluations on real-world medical datasets demonstrate the effectiveness of the proposed method in generating relational medical entity pairs that are meaningful and novel.", "target": ["Scoprire generativamente nuove coppie di entità significative con una certa relazione medica imparando puramente dalle coppie di entità significative esistenti, senza il requisito di un corpus di testo aggiuntivo per l'estrazione discriminante.", "Presenta un autoencoder variazionale per generare coppie di entità data una relazione in un ambiente medico.", "Nel contesto medico, questo articolo descrive il classico problema del \"completamento della knowledge base\" dai soli dati strutturati."]} +{"source": "Although variational autoencoders (VAEs) represent a widely influential deep generative model, many aspects of the underlying energy function remain poorly understood. In particular, it is commonly believed that Gaussian encoder/decoder assumptions reduce the effectiveness of VAEs in generating realistic samples . In this regard, we rigorously analyze the VAE objective, differentiating situations where this belief is and is not actually true . We then leverage the corresponding insights to develop a simple VAE enhancement that requires no additional hyperparameters or sensitive tuning . Quantitatively, this proposal produces crisp samples and stable FID scores that are actually competitive with a variety of GAN models, all while retaining desirable attributes of the original VAE architecture . The code for our model is available at \\url{https://github.com/daib13/TwoStageVAE}.", "target": ["Analizziamo da vicino la funzione obiettivo VAE e traiamo nuove conclusioni che portano a semplici miglioramenti.", "Propone un metodo VAE a due stadi per generare sample di alta qualità ed evitare la blurriness.", "Questo articolo analizza i VAE gaussiani.", "L'articolo fornisce una serie di risultati teorici sugli autoencoder variazionali gaussiani \"vanilla\", che sono poi utilizzati per costruire un nuovo algoritmo chiamato \"2 stage VAEs\"."]} +{"source": "We give a simple, fast algorithm for hyperparameter optimization inspired by techniques from the analysis of Boolean functions. We focus on the high-dimensional regime where the canonical example is training a neural network with a large number of hyperparameters. The algorithm --- an iterative application of compressed sensing techniques for orthogonal polynomials --- requires only uniform sampling of the hyperparameters and is thus easily parallelizable. Experiments for training deep neural networks on Cifar-10 show that compared to state-of-the-art tools (e.g., Hyperband and Spearmint), our algorithm finds significantly improved solutions, in some cases better than what is attainable by hand-tuning. In terms of overall running time (i.e., time required to sample various settings of hyperparameters plus additional computation time), we are at least an order of magnitude faster than Hyperband and Bayesian Optimization. We also outperform Random Search $8\\times$. Our method is inspired by provably-efficient algorithms for learning decision trees using the discrete Fourier transform. We obtain improved sample-complexty bounds for learning decision trees while matching state-of-the-art bounds on running time (polynomial and quasipolynomial, respectively).", "target": ["Un algoritmo di hyperparameter tuning utilizzando l'analisi di Fourier discreta e il rilevamento compresso", "Indaga il problema dell'ottimizzazione degli iperparametri sotto l'ipotesi che la funzione sconosciuta possa essere approssimata, mostrando che la minimizzazione approssimata può essere eseguita sull'ipercubo booleano.", "L'articolo esplora l'ottimizzazione degli iperparametri assumendo una struttura nella funzione sconosciuta che mappa gli iperparametri all'accuratezza della classificazione"]} +{"source": "Permutations and matchings are core building blocks in a variety of latent variable models, as they allow us to align, canonicalize, and sort data. Learning in such models is difficult, however, because exact marginalization over these combinatorial objects is intractable. In response, this paper introduces a collection of new methods for end-to-end learning in such models that approximate discrete maximum-weight matching using the continuous Sinkhorn operator. Sinkhorn iteration is attractive because it functions as a simple, easy-to-implement analog of the softmax operator. With this, we can define the Gumbel-Sinkhorn method, an extension of the Gumbel-Softmax method (Jang et al. 2016, Maddison2016 et al. 2016) to distributions over latent matchings. We demonstrate the effectiveness of our method by outperforming competitive baselines on a range of qualitatively different tasks: sorting numbers, solving jigsaw puzzles, and identifying neural signals in worms.", "target": ["Un nuovo metodo per l'inferenza delle permutazioni con gradient-descent, con applicazioni al latent matching inference e all'apprendimento supervisionato di permutazioni con reti neurali", "L'articolo utilizza un'approssimazione finita dell'operatore Sinkhorn per descrivere come si può costruire una rete neurale per l'apprendimento da dati di training valutati in permutazione.", "L'articolo propone un nuovo metodo che approssima il peso massimo discreto per l'apprendimento delle permutazioni latenti"]} +{"source": "Recent work in network quantization has substantially reduced the time and space complexity of neural network inference, enabling their deployment on embedded and mobile devices with limited computational and memory resources. However, existing quantization methods often represent all weights and activations with the same precision (bit-width). In this paper, we explore a new dimension of the design space: quantizing different layers with different bit-widths. We formulate this problem as a neural architecture search problem and propose a novel differentiable neural architecture search (DNAS) framework to efficiently explore its exponential search space with gradient-based optimization. Experiments show we surpass the state-of-the-art compression of ResNet on CIFAR-10 and ImageNet. Our quantized models with 21.1x smaller model size or 103.9x lower computational cost can still outperform baseline quantized or even full precision models.", "target": ["Un nuovo framework di ricerca dell'architettura neurale differenziabile per la quantizzazione mista di ConvNet.", "Gli autori introducono un nuovo metodo per la ricerca dell'architettura neurale che seleziona la quantizzazione della precisione dei pesi in ogni layer della rete neurale, e lo usano nel contesto della compressione della rete.", "L'articolo presenta un nuovo approccio nella quantizzazione della rete quantizzando diversi layer con diverse larghezze di bit e introduce un nuovo framework di ricerca dell'architettura neurale differenziabile."]} +{"source": "The top-$k$ error is a common measure of performance in machine learning and computer vision. In practice, top-$k$ classification is typically performed with deep neural networks trained with the cross-entropy loss. Theoretical results indeed suggest that cross-entropy is an optimal learning objective for such a task in the limit of infinite data. In the context of limited and noisy data however, the use of a loss function that is specifically designed for top-$k$ classification can bring significant improvements. Our empirical evidence suggests that the loss function must be smooth and have non-sparse gradients in order to work well with deep neural networks. Consequently, we introduce a family of smoothed loss functions that are suited to top-$k$ optimization via deep learning. The widely used cross-entropy is a special case of our family. Evaluating our smooth loss functions is computationally challenging: a na{\\\"i}ve algorithm would require $\\mathcal{O}(\\binom{n}{k})$ operations, where $n$ is the number of classes. Thanks to a connection to polynomial algebra and a divide-and-conquer approach, we provide an algorithm with a time complexity of $\\mathcal{O}(k n)$. Furthermore, we present a novel approximation to obtain fast and stable algorithms on GPUs with single floating point precision. We compare the performance of the cross-entropy loss and our margin-based losses in various regimes of noise and data size, for the predominant use case of $k=5$. Our investigation reveals that our loss is more robust to noise and overfitting than cross-entropy.", "target": ["Una loss smooth per la minimizzazione dell'errore Top-k", "Propone di utilizzare la loss top-k con deep model per affrontare il problema della confusione di classi con classi simili presenti o assenti nel dataset di training.", "Rende smooth le loss top-k.", "Questo articolo introduce una loss surrogata smooth per la SVM top-k, allo scopo di collegare gli SVM alle deep neural network."]} +{"source": "Designing a molecule with desired properties is one of the biggest challenges in drug development, as it requires optimization of chemical compound structures with respect to many complex properties. To augment the compound design process we introduce Mol-CycleGAN -- a CycleGAN-based model that generates optimized compounds with a chemical scaffold of interest. Namely, given a molecule our model generates a structurally similar one with an optimized value of the considered property. We evaluate the performance of the model on selected optimization objectives related to structural properties (presence of halogen groups, number of aromatic rings) and to a physicochemical property (penalized logP). In the task of optimization of penalized logP of drug-like molecules our model significantly outperforms previous results.", "target": ["Introduciamo Mol-CycleGAN - un nuovo modello generativo per l'ottimizzazione delle molecole per aiutare la progettazione dei farmaci.", "L'articolo presenta un approccio per l'ottimizzazione delle proprietà molecolari basato sull'applicazione di CycleGAN agli autoencoder variazionali per le molecole e impiega un VAE specifico del dominio chiamato Junction Tree VAE (JT-VAE).", "Questo articolo usa un autoencoder variazionale per imparare una funzione di traduzione, dall'insieme delle molecole senza la proprietà interessata all'insieme delle molecole con tale proprietà."]} +{"source": "Knowledge distillation is a potential solution for model compression. The idea is to make a small student network imitate the target of a large teacher network, then the student network can be competitive to the teacher one. Most previous studies focus on model distillation in the classification task, where they propose different architectures and initializations for the student network. However, only the classification task is not enough, and other related tasks such as regression and retrieval are barely considered. To solve the problem, in this paper, we take face recognition as a breaking point and propose model distillation with knowledge transfer from face classification to alignment and verification. By selecting appropriate initializations and targets in the knowledge transfer, the distillation can be easier in non-classification tasks. Experiments on the CelebA and CASIA-WebFace datasets demonstrate that the student network can be competitive to the teacher one in alignment and verification, and even surpasses the teacher network under specific compression rates. In addition, to achieve stronger knowledge transfer, we also use a common initialization trick to improve the distillation performance of classification. Evaluations on the CASIA-Webface and large-scale MS-Celeb-1M datasets show the effectiveness of this simple trick.", "target": ["Prendiamo il riconoscimento dei volti come punto di rottura e proponiamo la distillazione dei modelli con il transfer di conoscenze dalla classificazione dei volti all'allineamento e alla verifica", "Questo articolo propone di trasferire il classificatore dal modello per la classificazione dei volti al task di allineamento e verifica.", "Il manoscritto presenta esperimenti sulla knowledge distillation da un modello di classificazione dei volti a modelli di student per l'allineamento e la verifica dei volti."]} +{"source": "RNNs have been shown to be excellent models for sequential data and in particular for session-based user behavior. The use of RNNs provides impressive performance benefits over classical methods in session-based recommendations. In this work we introduce a novel ranking loss function tailored for RNNs in recommendation settings. The better performance of such loss over alternatives, along with further tricks and improvements described in this work, allow to achieve an overall improvement of up to 35% in terms of MRR and Recall@20 over previous session-based RNN solutions and up to 51% over classical collaborative filtering approaches. Unlike data augmentation-based improvements, our method does not increase training times significantly.", "target": ["Miglioramento delle raccomandazioni basate sulla sessione con RNN (GRU4Rec) del 35% utilizzando funzioni di loss e sampling di nuova concezione.", "Questo articolo analizza le funzioni di loss esistenti per le raccomandazioni basate sulla sessione e propone due nuove funzioni di loss che aggiungono una ponderazione alle funzioni di loss esistenti basate sul ranking", "Presenta modifiche sulla base di lavori precedenti per la raccomandazione basata sulla sessione usando RNN e pesando gli esempi negativi per la loro \"rilevanza\".", "Questo articolo discute i problemi di ottimizzazione delle funzioni di loss in GRU4Rec, propone dei trucchi per l'ottimizzazione e suggerisce una versione migliorata."]} +{"source": "In representational lifelong learning an agent aims to continually learn to solve novel tasks while updating its representation in light of previous tasks. Under the assumption that future tasks are related to previous tasks, representations should be learned in such a way that they capture the common structure across learned tasks, while allowing the learner sufficient flexibility to adapt to novel aspects of a new task. We develop a framework for lifelong learning in deep neural networks that is based on generalization bounds, developed within the PAC-Bayes framework. Learning takes place through the construction of a distribution over networks based on the tasks seen so far, and its utilization for learning a new task. Thus, prior knowledge is incorporated through setting a history-dependent prior for novel tasks. We develop a gradient-based algorithm implementing these ideas, based on minimizing an objective function motivated by generalization bounds, and demonstrate its effectiveness through numerical examples.", "target": ["Sviluppiamo un approccio di lifelong learning per transfer learning basato sulla teoria PAC-Bayes, in cui i prior vengono aggiustati quando si incontrano nuovi task, facilitando così l'apprendimento di nuovi task.", "Un nuovo bound sul rischio PAC-Bayesiano che serve come funzione obiettivo per l'apprendimento automatico multi-task, e un algoritmo per minimizzare una versione semplificata di quella funzione obiettivo.", "Estende i bound PAC-Bayes esistenti al multi-task learning, per consentire l'adattamento del prior a diversi task."]} +{"source": "Optimization algorithms for training deep models not only affects the convergence rate and stability of the training process, but are also highly related to the generalization performance of trained models. While adaptive algorithms, such as Adam and RMSprop, have shown better optimization performance than stochastic gradient descent (SGD) in many scenarios, they often lead to worse generalization performance than SGD, when used for training deep neural networks (DNNs). In this work, we identify two problems regarding the direction and step size for updating the weight vectors of hidden units, which may degrade the generalization performance of Adam. As a solution, we propose the normalized direction-preserving Adam (ND-Adam) algorithm, which controls the update direction and step size more precisely, and thus bridges the generalization gap between Adam and SGD. Following a similar rationale, we further improve the generalization performance in classification tasks by regularizing the softmax logits. By bridging the gap between SGD and Adam, we also shed some light on why certain optimization algorithms generalize better than others.", "target": ["Una versione su misura di Adam per il training delle DNN, che colma il divario di generalizzazione tra Adam e SGD.", "Propone una variante dell'algoritmo di ottimizzazione ADAM che normalizza i pesi di ogni hidden unit usando la batch normalization ", "Estensione dell'algoritmo di ottimizzazione di Adam per preservare la direzione di aggiornamento adattando il learning rate per i pesi in entrata a una hidden unit utilizzando congiuntamente la norma L2 del vettore del gradiente"]} +{"source": "Options in reinforcement learning allow agents to hierarchically decompose a task into subtasks, having the potential to speed up learning and planning. However, autonomously learning effective sets of options is still a major challenge in the field. In this paper we focus on the recently introduced idea of using representation learning methods to guide the option discovery process. Specifically, we look at eigenoptions, options obtained from representations that encode diffusive information flow in the environment. We extend the existing algorithms for eigenoption discovery to settings with stochastic transitions and in which handcrafted features are not available. We propose an algorithm that discovers eigenoptions while learning non-linear state representations from raw pixels. It exploits recent successes in the deep reinforcement learning literature and the equivalence between proto-value functions and the successor representation. We use traditional tabular domains to provide intuition about our approach and Atari 2600 games to demonstrate its potential.", "target": ["Mostriamo come possiamo usare la rappresentazione del successore per scoprire eigenoption in domini stocastici, a partire da pixel grezzi. Le eigenoption sono opzioni apprese per navigare nelle dimensioni latenti di una rappresentazione appresa.", "Estende l'idea di eigenoption a domini con transizioni stocastiche e dove le feature dello stato vengono apprese.", "Mostra l'equivalenza tra le proto value function e le rappresentazioni dei successori e deriva l'idea delle eigenoption come meccanismo nella scoperta delle opzioni", "L'articolo è un seguito del precedente lavoro di Machado et al. (2017) che mostra come le funzioni di proto-valore possono essere usate per definire opzioni chiamate \"eigenoption\"."]} +{"source": "One form of characterizing the expressiveness of a piecewise linear neural network is by the number of linear regions, or pieces, of the function modeled. We have observed substantial progress in this topic through lower and upper bounds on the maximum number of linear regions and a counting procedure. However, these bounds only account for the dimensions of the network and the exact counting may take a prohibitive amount of time, therefore making it infeasible to benchmark the expressiveness of networks. In this work, we approximate the number of linear regions of specific rectifier networks with an algorithm for probabilistic lower bounds of mixed-integer linear sets. In addition, we present a tighter upper bound that leverages network coefficients. We test both on trained networks. The algorithm for probabilistic lower bounds is several orders of magnitude faster than exact counting and the values reach similar orders of magnitude, hence making our approach a viable method to compare the expressiveness of such networks. The refined upper bound is particularly stronger on networks with narrow layers.", "target": ["Forniamo limiti superiori migliorati per il numero di regioni lineari usate nell'espressività della rete, e un algoritmo altamente efficiente (rispetto al conteggio esatto) per ottenere limiti inferiori probabilistici sul numero effettivo di regioni lineari.", "Contribuisce allo studio del numero di regioni lineari nelle reti neurali RELU utilizzando un algoritmo di conteggio probabilistico approssimato e l'analisi", "Costruisce il lavoro precedente che studia il conteggio delle regioni lineari nelle deep neural network, e migliora il limite superiore precedentemente proposto cambiando il vincolo di dimensionalità", "L'articolo si occupa dell'espressività di una rete neurale piecewise linear, caratterizzata dal numero di regioni lineari della funzione modellata, e sfrutta algoritmi probabilistici per calcolare i limiti più velocemente, e dimostra limiti più stretti."]} +{"source": "The ability to look multiple times through a series of pose-adjusted glimpses is fundamental to human vision. This critical faculty allows us to understand highly complex visual scenes. Short term memory plays an integral role in aggregating the information obtained from these glimpses and informing our interpretation of the scene. Computational models have attempted to address glimpsing and visual attention but have failed to incorporate the notion of memory. We introduce a novel, biologically inspired visual working memory architecture that we term the Hebb-Rosenblatt memory. We subsequently introduce a fully differentiable Short Term Attentive Working Memory model (STAWM) which uses transformational attention to learn a memory over each image it sees. The state of our Hebb-Rosenblatt memory is embedded in STAWM as the weights space of a layer. By projecting different queries through this layer we can obtain goal-oriented latent representations for tasks including classification and visual reconstruction. Our model obtains highly competitive classification performance on MNIST and CIFAR-10. As demonstrated through the CelebA dataset, to perform reconstruction the model learns to make a sequence of updates to a canvas which constitute a parts-based representation. Classification with the self supervised representation obtained from MNIST is shown to be in line with the state of the art models (none of which use a visual attention mechanism). Finally, we show that STAWM can be trained under the dual constraints of classification and reconstruction to provide an interpretable visual sketchpad which helps open the `black-box' of deep learning.", "target": ["Una working memory biologicamente ispirata che può essere integrata in modelli di attention visiva ricorrente per prestazioni allo stato dell'arte", "Introduce una nuova architettura di rete ispirata alla working memory visiva con attention e la applica a task di classificazione usandola come modello generativo", "L'articolo aumenta il modello di attention ricorrente con un nuovo modello di working memory di Hebb-Rosenblatt e raggiunge risultati competitivi su MNIST"]} +{"source": "Generative models have been successfully applied to image style transfer and domain translation. However, there is still a wide gap in the quality of results when learning such tasks on musical audio. Furthermore, most translation models only enable one-to-one or one-to-many transfer by relying on separate encoders or decoders and complex, computationally-heavy models. In this paper, we introduce the Modulated Variational auto-Encoders (MoVE) to perform musical timbre transfer. First, we define timbre transfer as applying parts of the auditory properties of a musical instrument onto another. We show that we can achieve and improve this task by conditioning existing domain translation techniques with Feature-wise Linear Modulation (FiLM). Then, by replacing the usual adversarial translation criterion by a Maximum Mean Discrepancy (MMD) objective, we alleviate the need for an auxiliary pair of discriminative networks. This allows a faster and more stable training, along with a controllable latent space encoder. By further conditioning our system on several different instruments, we can generalize to many-to-many transfer within a single variational architecture able to perform multi-domain transfers. Our models map inputs to 3-dimensional representations, successfully translating timbre from one instrument to another and supporting sound synthesis on a reduced set of control parameters. We evaluate our method in reconstruction and generation tasks while analyzing the auditory descriptor distributions across transferred domains. We show that this architecture incorporates generative controls in multi-domain transfer, yet remaining rather light, fast to train and effective on small datasets.", "target": ["L'articolo utilizza l'autoencoding variazionale e il condizionamento di rete per il transfer del timbro musicale, sviluppiamo e generalizziamo la nostra architettura per trasferimenti many-to-many tra strumenti insieme a visualizzazioni e valutazioni.", "Propone un auto-Encoder Variazionale Modulato per eseguire il transfer del timbro musicale sostituendo il solito criterio di traduzione avversaria con una Maximum Mean Discrepancy", "Descrive un modello many-to-many per il transfer del timbro musicale che si basa sui recenti sviluppi nel transfer del dominio e dello stile", "Propone un modello ibrido basato su VAE per eseguire il transfer del timbro su registrazioni di strumenti musicali."]} +{"source": "We study the behavior of weight-tied multilayer vanilla autoencoders under the assumption of random weights. Via an exact characterization in the limit of large dimensions, our analysis reveals interesting phase transition phenomena when the depth becomes large. This, in particular, provides quantitative answers and insights to three questions that were yet fully understood in the literature. Firstly, we provide a precise answer on how the random deep weight-tied autoencoder model performs “approximate inference” as posed by Scellier et al. (2018), and its connection to reversibility considered by several theoretical studies. Secondly, we show that deep autoencoders display a higher degree of sensitivity to perturbations in the parameters, distinct from the shallow counterparts. Thirdly, we obtain insights on pitfalls in training initialization practice, and demonstrate experimentally that it is possible to train a deep autoencoder, even with the tanh activation and a depth as large as 200 layers, without resorting to techniques such as layer-wise pre-training or batch normalization. Our analysis is not specific to any depths or any Lipschitz activations, and our analytical techniques may have broader applicability.", "target": ["Studiamo il comportamento degli autoencoder multilayer con pesi vincolati sotto l'ipotesi di pesi casuali. Attraverso una caratterizzazione esatta nel limite di grandi dimensioni, la nostra analisi rivela interessanti fenomeni di transizione di fase.", "Un'analisi teorica degli autoencoder con pesi vincolati tra encoder e decoder (weight-tied) tramite l'analisi del campo medio", "Analizza le prestazioni degli autoencoder con pesi vincolati basandosi sui recenti progressi nell'analisi dei problemi di statistica ad alta dimensione e in particolare, l'algoritmo di message passing", "Questo articolo studia gli autoencoder sotto diverse ipotesi, e sottolinea che questo modello di autoencoder casuale può essere analizzato in modo elegante e rigoroso con equazioni monodimensionali."]} +{"source": "Assessing distance betweeen the true and the sample distribution is a key component of many state of the art generative models, such as Wasserstein Autoencoder (WAE). Inspired by prior work on Sliced-Wasserstein Autoencoders (SWAE) and kernel smoothing we construct a new generative model – Cramer-Wold AutoEncoder (CWAE). CWAE cost function, based on introduced Cramer-Wold distance between samples, has a simple closed-form in the case of normal prior. As a consequence, while simplifying the optimization procedure (no need of sampling necessary to evaluate the distance function in the training loop), CWAE performance matches quantitatively and qualitatively that of WAE-MMD (WAE using maximum mean discrepancy based distance function) and often improves upon SWAE.", "target": ["Ispirati da lavori precedenti su Sliced-Wasserstein Autoencoders (SWAE) e kernel smoothing costruiamo un nuovo modello generativo - Cramer-Wold AutoEncoder (CWAE).", "Questo articolo propone una variante WAE basata su una nuova distanza statistica tra la distribuzione dei dati codificati e la distribuzione a priori latente", "Introduce una variazione dell'autoencoder di Wasserstein che è una nuova architettura di autoencoder regolarizzato che propone una scelta specifica della penalità di divergenza", "Questo articolo propone l'autoencoder Cramer-Wold, che utilizza la distanza Cramer-Wold tra due distribuzioni basata sul teorema Cramer-Wold."]} +{"source": "We propose a rejection sampling scheme using the discriminator of a GAN to approximately correct errors in the GAN generator distribution. We show that under quite strict assumptions, this will allow us to recover the data distribution exactly. We then examine where those strict assumptions break down and design a practical algorithm—called Discriminator Rejection Sampling (DRS)—that can be used on real data-sets. Finally, we demonstrate the efficacy of DRS on a mixture of Gaussians and on the state of the art SAGAN model. On ImageNet, we train an improved baseline that increases the best published Inception Score from 52.52 to 62.36 and reduces the Frechet Inception Distance from 18.65 to 14.79. We then use DRS to further improve on this baseline, improving the Inception Score to 76.08 and the FID to 13.75.", "target": ["Usiamo un discriminatore GAN per eseguire uno schema di rejection sampling approssimato sull'output del generatore GAN.", "Propone un algoritmo di rejection sampling per il sampling dal generatore GAN.", "Questo articolo ha proposto uno schema di rejection sampling post-elaborazione per le GAN, chiamato Discriminator Rejection Sampling, per aiutare a filtrare i sample \"buoni\" dal generatore delle GAN."]} +{"source": "The quality of the features used in visual recognition is of fundamental importance for the overall system. For a long time, low-level hand-designed feature algorithms as SIFT and HOG have obtained the best results on image recognition. Visual features have recently been extracted from trained convolutional neural networks. Despite the high-quality results, one of the main drawbacks of this approach, when compared with hand-designed features, is the training time required during the learning process. In this paper, we propose a simple and fast way to train supervised convolutional models to feature extraction while still maintaining its high-quality. This methodology is evaluated on different datasets and compared with state-of-the-art approaches.", "target": ["Un metodo semplice e veloce per estrarre le feature visive dalle reti neurali convoluzionali", "Propone un modo veloce per imparare feature convoluzionali che in seguito possono essere utilizzate con qualsiasi classificatore utilizzando un numero ridotto di epoch di allenamento e schedule delay del learning rate", "Utilizzare uno schema di learning rate decay che è fisso rispetto al numero di epoche utilizzate nel training ed estrarre l'output del penultimo layer come feature per addestrare un classificatore convenzionale."]} +{"source": "We develop a framework for understanding and improving recurrent neural networks (RNNs) using max-affine spline operators (MASOs). We prove that RNNs using piecewise affine and convex nonlinearities can be written as a simple piecewise affine spline operator. The resulting representation provides several new perspectives for analyzing RNNs, three of which we study in this paper. First, we show that an RNN internally partitions the input space during training and that it builds up the partition through time. Second, we show that the affine slope parameter of an RNN corresponds to an input-specific template, from which we can interpret an RNN as performing a simple template matching (matched filtering) given the input. Third, by carefully examining the MASO RNN affine mapping, we prove that using a random initial hidden state corresponds to an explicit L2 regularization of the affine parameters, which can mollify exploding gradients and improve generalization. Extensive experiments on several datasets of various modalities demonstrate and validate each of the above conclusions. In particular, using a random initial hidden states elevates simple RNNs to near state-of-the-art performers on these datasets.", "target": ["Forniamo nuove intuizioni e interpretazioni delle RNN dal punto di vista degli operatori spline max-affine.", "Riscrive le equazioni di Elman RNN in termini dei cosiddetti operatori spline max-affine ", "Fornire un nuovo approccio verso la comprensione delle RNN utilizzando operatori spline max-affline (MASO) riscrivendoli con attivazioni affini e convesse piecewise MASO", "Gli autori si basano sull'interpetazione dell'operatore spline max-affine di una classe sostanziale di reti profonde, concentrandosi sulle reti neurali ricorrenti che utilizzano il rumore nell'hidden state iniziale come regolarizzazione"]} +{"source": "Reasoning over text and Knowledge Bases (KBs) is a major challenge for Artificial Intelligence, with applications in machine reading, dialogue, and question answering. Transducing text to logical forms which can be operated on is a brittle and error-prone process . Operating directly on text by jointly learning representations and transformations thereof by means of neural architectures that lack the ability to learn and exploit general rules can be very data-inefficient and not generalise correctly . These issues are addressed by Neural Theorem Provers (NTPs) (Rocktäschel & Riedel, 2017), neuro-symbolic systems based on a continuous relaxation of Prolog’s backward chaining algorithm, where symbolic unification between atoms is replaced by a differentiable operator computing the similarity between their embedding representations . In this paper, we first propose Neighbourhood-approximated Neural Theorem Provers (NaNTPs) consisting of two extensions toNTPs, namely a) a method for drastically reducing the previously prohibitive time and space complexity during inference and learning, and b) an attention mechanism for improving the rule learning process, deeming them usable on real-world datasets. Then, we propose a novel approach for jointly reasoning over KB facts and textual mentions, by jointly embedding them in a shared embedding space. The proposed method is able to extract rules and provide explanations—involving both textual patterns and KB relations—from large KBs and text corpora. We show that NaNTPs perform on par with NTPs at a fraction of a cost, and can achieve competitive link prediction results on challenging large-scale datasets, including WN18, WN18RR, and FB15k-237 (with and without textual mentions) while being able to provide explanations for each prediction and extract interpretable rules.", "target": ["Scaliamo i Neural Theorem Prover a grandi insiemi di dati, miglioriamo il processo di apprendimento delle regole e lo estendiamo per ragionare insieme su testo e knowledge base.", "Propone un'estensione del sistema Neural Theorem Provers che affronta i problemi principali di questo modello riducendo la complessità temporale e spaziale del modello", "Scala gli NTP usando la ricerca approssimativa dei nearest neighbour su fatti e regole durante l'unificazione e suggerisce di parametrizzare i predicati usando l'attention su predicati conosciuti", "Migliora l'approccio Neural Theorem Prover proposto in precedenza utilizzando la nearest neighbor search."]} +{"source": "We investigate the methods by which a Reservoir Computing Network (RCN) learns concepts such as 'similar' and 'different' between pairs of images using a small training dataset and generalizes these concepts to previously unseen types of data. Specifically, we show that an RCN trained to identify relationships between image-pairs drawn from a subset of digits from the MNIST database or the depth maps of subset of visual scenes from a moving camera generalizes the learned transformations to images of digits unseen during training or depth maps of different visual scenes. We infer, using Principal Component Analysis, that the high dimensional reservoir states generated from an input image pair with a specific transformation converge over time to a unique relationship. Thus, as opposed to training the entire high dimensional reservoir state, the reservoir only needs to train on these unique relationships, allowing the reservoir to perform well with very few training examples. Thus, generalization of learning to unseen images is interpretable in terms of clustering of the reservoir state onto the attractor corresponding to the transformation in reservoir space. We find that RCNs can identify and generalize linear and non-linear transformations, and combinations of transformations, naturally and be a robust and effective image classifier. Additionally, RCNs perform significantly better than state of the art neural network classification techniques such as deep Siamese Neural Networks (SNNs) in generalization tasks both on the MNIST dataset and more complex depth maps of visual scenes from a moving camera. This work helps bridge the gap between explainable machine learning and biological learning through analogies using small datasets, and points to new directions in the investigation of learning processes.", "target": ["Generalizzazione delle relazioni apprese tra coppie di immagini utilizzando una piccola quantità di dati di training a tipi di immagini precedentemente non visti utilizzando un modello di sistemi dinamici spiegabili, Reservoir Computing, e una tecnica di apprendimento biologicamente plausibile basata sulle analogie.", "Propone dei risultati di \"combining transformations\" nel contesto di RC utilizzando una echo-state network con accattivazioni tanh standard con la differenza che i pesi ricorrenti non sono addestrati", "Nuovo metodo per classificare diverse distorsioni dei dati MNIST", "L'articolo utilizza una echo-state network per imparare a classificare le trasformazioni di immagini tra coppie di immagini in una delle cinque classi."]} +{"source": "We present Generative Adversarial Privacy and Fairness (GAPF), a data-driven framework for learning private and fair representations of the data. GAPF leverages recent advances in adversarial learning to allow a data holder to learn \"universal\" representations that decouple a set of sensitive attributes from the rest of the dataset. Under GAPF, finding the optimal decorrelation scheme is formulated as a constrained minimax game between a generative decorrelator and an adversary. We show that for appropriately chosen adversarial loss functions, GAPF provides privacy guarantees against strong information-theoretic adversaries and enforces demographic parity. We also evaluate the performance of GAPF on multi-dimensional Gaussian mixture models and real datasets, and show how a designer can certify that representations learned under an adversary with a fixed architecture perform well against more complex adversaries.", "target": ["Presentiamo Generative Adversarial Privacy and Fairness (GAPF), un framework data-driven per l'apprendimento di rappresentazioni private e fair con garanzie certificate di privacy/fairness", "Questo articolo usa un modello GAN per fornire una panoramica del lavoro correlato all'apprendimento di rappresentazioni private/fair (PRL).", "Questo articolo presenta un approccio adversarial per rappresentazioni private e fair attraverso la distorsione appresa dei dati che minimizza la dipendenza dalle variabili sensibili mentre il grado di distorsione è vincolato.", "Gli autori descrivono un framework relativo all'imparare una rappresentazione di parità demografica che può essere utilizzata per addestrare alcuni classificatori."]} +{"source": "Current machine learning algorithms can be easily fooled by adversarial examples. One possible solution path is to make models that use confidence thresholding to avoid making mistakes. Such models refuse to make a prediction when they are not confident of their answer. We propose to evaluate such models in terms of tradeoff curves with the goal of high success rate on clean examples and low failure rate on adversarial examples. Existing untargeted attacks developed for models that do not use confidence thresholding tend to underestimate such models' vulnerability. We propose the MaxConfidence family of attacks, which are optimal in a variety of theoretical settings, including one realistic setting: attacks against linear models. Experiments show the attack attains good results in practice. We show that simple defenses are able to perform well on MNIST but not on CIFAR, contributing further to previous calls that MNIST should be retired as a benchmarking dataset for adversarial robustness research. We release code for these evaluations as part of the cleverhans (Papernot et al 2018) library (ICLR reviewers should be careful not to look at who contributed these features to cleverhans to avoid de-anonymizing this submission).", "target": ["Presentiamo una metrica e un attacco ottimale per valutare i modelli che si difendono dagli adversarial example usando confidence thresholding", "Questo articolo introduce una famiglia di attacchi agli algoritmi di confidence thresholding, concentrandosi principalmente sulle metodologie di valutazione.", "Propone un metodo di valutazione per i modelli di difesa di confidence thresholding e un approccio per generare adversarial example scegliendo la classe sbagliata con più fiducia quando si usano attacchi mirati", "L'articolo presenta una metodologia di valutazione degli attacchi ai metodi di confidence thresholding e propone un nuovo tipo di attacco."]} +{"source": "Deep learning has achieved remarkable successes in solving challenging reinforcement learning (RL) problems when dense reward function is provided. However, in sparse reward environment it still often suffers from the need to carefully shape reward function to guide policy optimization. This limits the applicability of RL in the real world since both reinforcement learning and domain-specific knowledge are required. It is therefore of great practical importance to develop algorithms which can learn from a binary signal indicating successful task completion or other unshaped, sparse reward signals. We propose a novel method called competitive experience replay, which efficiently supplements a sparse reward by placing learning in the context of an exploration competition between a pair of agents. Our method complements the recently proposed hindsight experience replay (HER) by inducing an automatic exploratory curriculum. We evaluate our approach on the tasks of reaching various goal locations in an ant maze and manipulating objects with a robotic arm. Each task provides only binary rewards indicating whether or not the goal is achieved. Our method asymmetrically augments these sparse rewards for a pair of agents each learning the same task, creating a competitive game designed to drive exploration. Extensive experiments demonstrate that this method leads to faster converge and improved task performance.", "target": ["Un nuovo metodo per imparare con reward sparse usando l'adversarial re-labeling della reward", "Propone di utilizzare un ambiente competitivo multi-agente per incoraggiare l'esplorazione e mostra che CER + HER > HER ~ CER", "Proporre un nuovo metodo per imparare da reward sparse in setting di reinforcement learning model-free e rendendo la reward densa", "Per affrontare i problemi di reward sparsa e incoraggiare l'esplorazione negli algoritmi RL, gli autori propongono una strategia di relabeling chiamata Competitive Experience Reply (CER)."]} +{"source": "This paper proposes a neural end-to-end text-to-speech (TTS) model which can control latent attributes in the generated speech that are rarely annotated in the training data, such as speaking style, accent, background noise, and recording conditions. The model is formulated as a conditional generative model with two levels of hierarchical latent variables. The first level is a categorical variable, which represents attribute groups (e.g. clean/noisy) and provides interpretability. The second level, conditioned on the first, is a multivariate Gaussian variable, which characterizes specific attribute configurations (e.g. noise level, speaking rate) and enables disentangled fine-grained control over these attributes. This amounts to using a Gaussian mixture model (GMM) for the latent distribution. Extensive evaluation demonstrates its ability to control the aforementioned attributes. In particular, it is capable of consistently synthesizing high-quality clean speech regardless of the quality of the training data for the target speaker.", "target": ["Costruire un modello TTS con Gaussian Mixture VAE permette un controllo preciso dello stile di conversazione, delle condizioni di rumore e altro.", "Descrive il modello GAN condizionato per generare gli spettri di Mel condizionati dal parlante aumentando lo spazio z corrispondente all'identificazione", "Questo articolo propone un modello con variabili latenti a due layer per ottenere una rappresentazione latente disentangled, facilitando così un preciso controllo su vari attributi", "Questo articolo propone un modello che può controllare gli attributi non annotati, come lo stile di conversazione, l'accento, il rumore di fondo, ecc."]} +{"source": "Visual Question Answering (VQA) models have struggled with counting objects in natural images so far. We identify a fundamental problem due to soft attention in these models as a cause. To circumvent this problem, we propose a neural network component that allows robust counting from object proposals. Experiments on a toy task show the effectiveness of this component and we obtain state-of-the-art accuracy on the number category of the VQA v2 dataset without negatively affecting other categories, even outperforming ensemble models with our single model. On a difficult balanced pair metric, the component gives a substantial improvement in counting over a strong baseline by 6.6%.", "target": ["Abilitare i modelli di Visual Question Answering a contare gestendo le proposte di oggetti sovrapposti.", "Questo articolo propone un'architettura di rete progettata a mano su un grafo di proposte di oggetti per eseguire una non-maximum suppression per ottenere il conteggio degli oggetti.", "Si concentra su un problema di conteggio nel visual question answering utilizzando il meccanismo dell'attention e propone un componente di conteggio differenziabile che conta esplicitamente il numero di oggetti", "Questo articolo affronta il problema del conteggio degli oggetti nella visual question answering, propone molte euristiche per trovare il conteggio corretto."]} +{"source": "We propose a simple and robust training-free approach for building sentence representations. Inspired by the Gram-Schmidt Process in geometric theory, we build an orthogonal basis of the subspace spanned by a word and its surrounding context in a sentence. We model the semantic meaning of a word in a sentence based on two aspects. One is its relatedness to the word vector subspace already spanned by its contextual words. The other is its novel semantic meaning which shall be introduced as a new basis vector perpendicular to this existing subspace. Following this motivation, we develop an innovative method based on orthogonal basis to combine pre-trained word embeddings into sentence representation. This approach requires zero training and zero parameters, along with efficient inference performance. We evaluate our approach on 11 downstream NLP tasks. Experimental results show that our model outperforms all existing zero-training alternatives in all the tasks and it is competitive to other approaches relying on either large amounts of labelled data or prolonged training time.", "target": ["Un approccio semplice e senza training per l'embedding delle frasi con prestazioni competitive rispetto ai modelli sofisticati che richiedono una grande quantità di dati di training o un tempo di training prolungato.", "Presentato un nuovo modo senza training di generare sentence embedding con un'analisi sistematica", "Propone un nuovo metodo basato sulla geometria per l'embedding delle frasi dai word embedding, quantificando la novità, il significato e l'unicità nel corpus di ogni parola", "Questo articolo esplora l'embedding delle frasi basato sulla decomposizione ortogonale dello spazio coperto dai word embedding"]} +{"source": "In few-shot classification, we are interested in learning algorithms that train a classifier from only a handful of labeled examples. Recent progress in few-shot classification has featured meta-learning, in which a parameterized model for a learning algorithm is defined and trained on episodes representing different classification problems, each with a small labeled training set and its corresponding test set. In this work, we advance this few-shot classification paradigm towards a scenario where unlabeled examples are also available within each episode. We consider two situations: one where all unlabeled examples are assumed to belong to the same set of classes as the labeled examples of the episode, as well as the more challenging situation where examples from other distractor classes are also provided. To address this paradigm, we propose novel extensions of Prototypical Networks (Snell et al., 2017) that are augmented with the ability to use unlabeled examples when producing prototypes. These models are trained in an end-to-end way on episodes, to learn to leverage the unlabeled examples successfully. We evaluate these methods on versions of the Omniglot and miniImageNet benchmarks, adapted to this new framework augmented with unlabeled examples. We also propose a new split of ImageNet, consisting of a large set of classes, with a hierarchical structure. Our experiments confirm that our Prototypical Networks can learn to improve their predictions due to unlabeled examples, much like a semi-supervised algorithm would.", "target": ["Proponiamo nuove estensioni delle Prototypical Network che sono aumentate con la capacità di utilizzare esempi non annotati quando si producono prototipi.", "Questo documento è un'estensione di una Prototypical Network che considera l'impiego degli esempi non annotati disponibili per aiutare il training in ogni episodio", "Studia il problema della classificazione semi-supervisionata few-shot estendendo le reti prototipiche al setting dell'apprendimento semi-supervised con esempi di classi distrattrici", "Estende la Prototypical Network al setting semi-supervisionato aggiornando i prototipi usando pseudo-label assegnate, trattando i distrattori e pesando i sample usando la distanza dai prototipi originali."]} +{"source": "We investigate the properties of multidimensional probability distributions in the context of latent space prior distributions of implicit generative models. Our work revolves around the phenomena arising while decoding linear interpolations between two random latent vectors -- regions of latent space in close proximity to the origin of the space are oversampled, which restricts the usability of linear interpolations as a tool to analyse the latent space. We show that the distribution mismatch can be eliminated completely by a proper choice of the latent probability distribution or using non-linear interpolations. We prove that there is a trade off between the interpolation being linear, and the latent distribution having even the most basic properties required for stable training, such as finite mean. We use the multidimensional Cauchy distribution as an example of the prior distribution, and also provide a general method of creating non-linear interpolations, that is easily applicable to a large family of commonly used latent distributions.", "target": ["Proviamo teoricamente che le interpolazioni lineari sono inadatte all'analisi dei modelli generativi impliciti addestrati.", "Studia il problema di quando l'interpolante lineare tra due variabili casuali segue la stessa distribuzione, in relazione alla distribuzione a priori di un modello generativo implicito", "Questo lavoro chiede come interpolare nello spazio latente dato un modello di variabile latente."]} +{"source": "Deep neural networks (DNN) have shown promising performance in computer vision. In medical imaging, encouraging results have been achieved with deep learning for applications such as segmentation, lesion detection and classification. Nearly all of the deep learning based image analysis methods work on reconstructed images, which are obtained from original acquisitions via solving inverse problems (reconstruction). The reconstruction algorithms are designed for human observers, but not necessarily optimized for DNNs which can often observe features that are incomprehensible for human eyes. Hence, it is desirable to train the DNNs directly from the original data which lie in a different domain with the images. In this paper, we proposed an end-to-end DNN for abnormality detection in medical imaging. To align the acquisition with the annotations made by radiologists in the image domain, a DNN was built as the unrolled version of iterative reconstruction algorithms to map the acquisitions to images, and followed by a 3D convolutional neural network (CNN) to detect the abnormality in the reconstructed images. The two networks were trained jointly in order to optimize the entire DNN for the detection task from the original acquisitions. The DNN was implemented for lung nodule detection in low-dose chest computed tomography (CT), where a numerical simulation was done to generate acquisitions from 1,018 chest CT images with radiologists' annotations. The proposed end-to-end DNN demonstrated better sensitivity and accuracy for the task compared to a two-step approach, in which the reconstruction and detection DNNs were trained separately. A significant reduction of false positive rate on suspicious lesions were observed, which is crucial for the known over-diagnosis in low-dose lung CT imaging. The images reconstructed by the proposed end-to-end network also presented enhanced details in the region of interest.", "target": ["Rilevamento del nodulo polmonare a partire da dati di proiezione piuttosto che da immagini.", "Le DNN sono utilizzate per il rilevamento dei noduli polmonari basato su patch nei dati di proiezione CT.", "Modellazione congiunta della ricostruzione della tomografia computerizzata e del rilevamento delle lesioni nel polmone addestrando la mappatura dal sinogramma grezzo agli output di rilevamento in modo end-to-end", "Presenta un training end-to-end di un'architettura CNN che combina l'elaborazione del segnale dell'immagine CT e l'analisi dell'immagine."]} +{"source": "Deep reinforcement learning (DRL) algorithms have demonstrated progress in learning to find a goal in challenging environments. As the title of the paper by Mirowski et al. (2016) suggests, one might assume that DRL-based algorithms are able to “learn to navigate” and are thus ready to replace classical mapping and path-planning algorithms, at least in simulated environments. Yet, from experiments and analysis in this earlier work, it is not clear what strategies are used by these algorithms in navigating the mazes and finding the goal. In this paper, we pose and study this underlying question: are DRL algorithms doing some form of mapping and/or path-planning? Our experiments show that the algorithms are not memorizing the maps of mazes at the testing stage but, rather, at the training stage. Hence, the DRL algorithms fall short of qualifying as mapping or path-planning algorithms with any reasonable definition of mapping. We extend the experiments in Mirowski et al. (2016) by separating the set of training and testing maps and by a more ablative coverage of the space of experiments. Our systematic experiments show that the NavA3C-D1-D2-L algorithm, when trained and tested on the same maps, is able to choose the shorter paths to the goal. However, when tested on unseen maps the algorithm utilizes a wall-following strategy to find the goal without doing any mapping or path planning.", "target": ["Valutiamo quantitativamente e qualitativamente i metodi di navigazione basati sul deep reinforcement learning in una varietà di condizioni per rispondere alla domanda su quanto siano vicini a sostituire i pianificatori di percorso classici e gli algoritmi di mapping.", "Valutare un modello basato su Deep RL su labirinti di allenamento misurando la repeated latency to goal e il confronto con il percorso più breve"]} +{"source": "In many robotic applications, it is crucial to maintain a belief about the state of a system, like the location of a robot or the pose of an object. These state estimates serve as input for planning and decision making and provide feedback during task execution. Recursive Bayesian Filtering algorithms address the state estimation problem, but they require a model of the process dynamics and the sensory observations as well as noise estimates that quantify the accuracy of these models. Recently, multiple works have demonstrated that the process and sensor models can be learned by end-to-end training through differentiable versions of Recursive Filtering methods. However, even if the predictive models are known, finding suitable noise models remains challenging. Therefore, many practical applications rely on very simplistic noise models. Our hypothesis is that end-to-end training through differentiable Bayesian Filters enables us to learn more complex heteroscedastic noise models for the system dynamics. We evaluate learning such models with different types of filtering algorithms and on two different robotic tasks. Our experiments show that especially for sampling-based filters like the Particle Filter, learning heteroscedastic noise models can drastically improve the tracking performance in comparison to using constant noise models.", "target": ["Valutiamo l'apprendimento di modelli di rumore eteroschedastici all'interno di diversi filtri di Bayes differenziabili", "Propone di apprendere modelli di rumore eteroscedastici dai dati ottimizzando la probabilità di predizione end-to-end attraverso filtri bayesiani differenziabili e due diverse versioni del filtro di Kalman unscented", "Rivede i filtri di Bayes e valuta il beneficio di addestrare i modelli di osservazione e di rumore di processo mantenendo fissi tutti gli altri modelli", "Questo articolo presenta un metodo per imparare e utilizzare il rumore dipendente dallo stato e dall'osservazione nei tradizionali algoritmi di filtraggio bayesiano. L'approccio consiste nel costruire un modello di rete neurale che prende come input i dati grezzi di osservazione e produce una rappresentazione compatta e una covarianza diagonale associata."]} +{"source": "Graph convolutional neural networks have recently shown great potential for the task of zero-shot learning. These models are highly sample efficient as related concepts in the graph structure share statistical strength allowing generalization to new classes when faced with a lack of data. However, we find that the extensive use of Laplacian smoothing at each layer in current approaches can easily dilute the knowledge from distant nodes and consequently decrease the performance in zero-shot learning. In order to still enjoy the benefit brought by the graph structure while preventing the dilution of knowledge from distant nodes, we propose a Dense Graph Propagation (DGP) module with carefully designed direct links among distant nodes. DGP allows us to exploit the hierarchical graph structure of the knowledge graph through additional connections. These connections are added based on a node's relationship to its ancestors and descendants. A weighting scheme is further used to weigh their contribution depending on the distance to the node. Combined with finetuning of the representations in a two-stage training approach our method outperforms state-of-the-art zero-shot learning approaches.", "target": ["Ripensiamo al modo in cui le informazioni possono essere sfruttate in modo più efficiente nel grafo della conoscenza per migliorare le prestazioni nel task di apprendimento Zero-Shot e proponiamo un modulo DGP (dense graph propagation) per questo scopo.", "Questi autori propongono una soluzione al problema dell'over-smoothing nelle graph convolutional neural network permettendo una propagazione densa tra tutti i nodi correlati, ponderata dalla distanza reciproca.", "Propone una nuova graph convolutional neural network per affrontare il problema della classificazione zero-shot utilizzando strutture relazionali tra le classi come input delle graph convolutional neural network per imparare classificatori per classi non viste"]} +{"source": "In this paper, we propose a capsule-based neural network model to solve the semantic segmentation problem. By taking advantage of the extractable part-whole dependencies available in capsule layers, we derive the probabilities of the class labels for individual capsules through a recursive, layer-by-layer procedure. We model this procedure as a traceback pipeline and take it as a central piece to build an end-to-end segmentation network. Under the proposed framework, image-level class labels and object boundaries are jointly sought in an explicit manner, which poses a significant advantage over the state-of-the-art fully convolutional network (FCN) solutions. Experiments conducted on modified MNIST and neuroimages demonstrate that our model considerably enhance the segmentation performance compared to the leading FCN variant.", "target": ["Una segmentazione semantica basata su capsule, in cui le probabilità delle label di classe sono rintracciate attraverso la pipeline di capsule.", "Gli autori presentano un meccanismo di tracciamento per associare il livello più basso di Capsule con le loro rispettive classi", "Propone un layer di tracciamento per capsule network per fare la segmentazione semantica e fa uso esplicito della relazione parte-intero nei capsule layer", "Propone un metodo trace-back basato sul concetto CapsNet di Sabour per eseguire una segmentazione semantica in parallelo alla classificazione."]} +{"source": "Studying the evolution of information theoretic quantities during Stochastic Gradient Descent (SGD) learning of Artificial Neural Networks (ANNs) has gained popularity in recent years. Nevertheless, these type of experiments require estimating mutual information and entropy which becomes intractable for moderately large problems. In this work we propose a framework for understanding SGD learning in the information plane which consists of observing entropy and conditional entropy of the output labels of ANN. Through experimental results and theoretical justifications it is shown that, under some assumptions, the SGD learning trajectories appear to be similar for different ANN architectures. First, the SGD learning is modeled as a Hidden Markov Process (HMP) whose entropy tends to increase to the maximum. Then, it is shown that the SGD learning trajectory appears to move close to the shortest path between the initial and final joint distributions in the space of probability measures equipped with the total variation metric. Furthermore, it is shown that the trajectory of learning in the information plane can provide an alternative for observing the learning process, with potentially richer information about the learning than the trajectories in training and test error.", "target": ["Consideriamo SGD come una traiettoria nello spazio delle misure di probabilità, mostriamo la sua connessione con i processi di Markov, proponiamo un semplice modello di Markov per l'apprendimento di SGD e lo confrontiamo sperimentalmente con SGD utilizzando quantità dalla teoria dell'informazione.", "Costruisce una catena di Markov che segue un percorso corto nella metrica TV su P e mostra che le traiettorie di SGD e \\alpha-SMLC hanno entropia condizionale simile", "Studia la traiettoria di H(\\hat{y}) contro H(\\hat{y}|y) sul piano dell'informazione per i metodi di stochastic gradient descent per il training delle reti neurali", "Descrive SGD dal punto di vista della distribuzione p(y',y) dove y è una vera label di classe (eventualmente corrotta) e y' una predizione del modello."]} +{"source": "Stochastic gradient Markov chain Monte Carlo (SG-MCMC) has become increasingly popular for simulating posterior samples in large-scale Bayesian modeling. However, existing SG-MCMC schemes are not tailored to any specific probabilistic model, even a simple modification of the underlying dynamical system requires significant physical intuition. This paper presents the first meta-learning algorithm that allows automated design for the underlying continuous dynamics of an SG-MCMC sampler. The learned sampler generalizes Hamiltonian dynamics with state-dependent drift and diffusion, enabling fast traversal and efficient exploration of energy landscapes. Experiments validate the proposed approach on Bayesian fully connected neural network, Bayesian convolutional neural network and Bayesian recurrent neural network tasks, showing that the learned sampler outperforms generic, hand-designed SG-MCMC algorithms, and generalizes to different datasets and larger architectures.", "target": ["Questo articolo propone un metodo per automatizzare la progettazione della proposta MCMC a gradiente stocastico usando un approccio di meta learning.", "Propone un approccio di meta-learning per progettare automaticamente un sampler MCMC basato sulla dinamica hamiltoniana per fare mixin più velocemente su problemi simili a quelli di allenamento", "Parametrizza le matrici di diffusione e di curl tramite reti neurali e meta-apprende e ottimizza un algoritmo sg-mcmc."]} +{"source": "We propose a new, multi-component energy function for energy-based Generative Adversarial Networks (GANs) based on methods from the image quality assessment literature. Our approach expands on the Boundary Equilibrium Generative Adversarial Network (BEGAN) by outlining some of the short-comings of the original energy and loss functions. We address these short-comings by incorporating an l1 score, the Gradient Magnitude Similarity score, and a chrominance score into the new energy function. We then provide a set of systematic experiments that explore its hyper-parameters. We show that each of the energy function's components is able to represent a slightly different set of features, which require their own evaluation criteria to assess whether they have been adequately learned. We show that models using the new energy function are able to produce better image representations than the BEGAN model in predicted ways.", "target": ["Le tecniche di valutazione della qualità dell'immagine migliorano il training e la valutazione delle generative adversarial network basate sull'energia", "Propone una formulazione basata sull'energia per il modello BEGAN e la modifica per includere un termine basato sulla valutazione della qualità dell'immagine", "Propone alcune nuove funzioni energetiche nel BEGAN (boundary equilibrium GAN framework), tra cui l_1 score, Gradient magnitude similarity score, e chrominance score."]} +{"source": "Momentum is a simple and widely used trick which allows gradient-based optimizers to pick up speed along low curvature directions. Its performance depends crucially on a damping coefficient. Largecamping coefficients can potentially deliver much larger speedups, but are prone to oscillations and instability; hence one typically resorts to small values such as 0.5 or 0.9. We propose Aggregated Momentum (AggMo), a variant of momentum which combines multiple velocity vectors with different damping coefficients. AggMo is trivial to implement, but significantly dampens oscillations, enabling it to remain stable even for aggressive damping coefficients such as 0.999. We reinterpret Nesterov's accelerated gradient descent as a special case of AggMo and analyze rates of convergence for quadratic objectives. Empirically, we find that AggMo is a suitable drop-in replacement for other momentum methods, and frequently delivers faster convergence with little to no tuning.", "target": ["Introduciamo una semplice variante dell'ottimizzazione del momentum che è in grado di superare il momentum classico, Nesterov e Adam su task di deep learning con una sintonizzazione minima degli iperparametri.", "Introduce una variante del momentum che aggrega diverse velocità con diversi coefficienti di smorzamento che diminuisce significativamente l'oscillazione", "Proposto un metodo di momentum aggregato per l'ottimizzazione basata sul gradiente utilizzando più vettori di velocità con diversi fattori di smorzamento invece di un singolo vettore di velocità per migliorare la stabilità", "Gli autori combinano diversi update step insieme per ottenere il momentum aggregato dimostrando anche che è più stabile degli altri metodi di momentum"]} +{"source": "Recurrent Neural Networks architectures excel at processing sequences by modelling dependencies over different timescales. The recently introduced Recurrent Weighted Average (RWA) unit captures long term dependencies far better than an LSTM on several challenging tasks. The RWA achieves this by applying attention to each input and computing a weighted average over the full history of its computations. Unfortunately, the RWA cannot change the attention it has assigned to previous timesteps, and so struggles with carrying out consecutive tasks or tasks with changing requirements. We present the Recurrent Discounted Attention (RDA) unit that builds on the RWA by additionally allowing the discounting of the past. We empirically compare our model to RWA, LSTM and GRU units on several challenging tasks. On tasks with a single output the RWA, RDA and GRU units learn much quicker than the LSTM and with better performance. On the multiple sequence copy task our RDA unit learns the task three times as quickly as the LSTM or GRU units while the RWA fails to learn at all. On the Wikipedia character prediction task the LSTM performs best but it followed closely by our RDA unit. Overall our RDA unit performs well and is sample efficient on a large variety of sequence tasks.", "target": ["Introduciamo la Recurrent Discounted Unit che applica l'attention a qualsiasi lunghezza di sequenza in tempo lineare", "Questo articolo propone la Recurrent Discounted Attention (RDA), un'estensione della Recurrent Weighted Average (RWA) aggiungendo un fattore di sconto.", "Estende la media dei pesi ricorrenti per superare la limitazione del metodo originale mantenendo il suo vantaggio e propone il metodo che usa le reti di Elman come RNN di base"]} +{"source": "Ordinary stochastic neural networks mostly rely on the expected values of their weights to make predictions, whereas the induced noise is mostly used to capture the uncertainty, prevent overfitting and slightly boost the performance through test-time averaging. In this paper, we introduce variance layers, a different kind of stochastic layers. Each weight of a variance layer follows a zero-mean distribution and is only parameterized by its variance. It means that each object is represented by a zero-mean distribution in the space of the activations. We show that such layers can learn surprisingly well, can serve as an efficient exploration tool in reinforcement learning tasks and provide a decent defense against adversarial attacks. We also show that a number of conventional Bayesian neural networks naturally converge to such zero-mean posteriors. We observe that in these cases such zero-mean parameterization leads to a much better training objective than more flexible conventional parameterizations where the mean is being learned.", "target": ["È possibile imparare una distribuzione gaussiana centrata sullo zero sui pesi di una rete neurale imparando solo le varianze, e funziona sorprendentemente bene.", "Questo articolo studia gli effetti della media del posterior variazionale e propone il layer della varianza, che usa solo la varianza per memorizzare le informazioni", "Studia le reti neurali a varianza che approssimano il posterior delle reti neurali bayesiane con distribuzioni gaussiane a media zero"]} +{"source": "Graph Convolutional Networks (GCNs) are a recently proposed architecture which has had success in semi-supervised learning on graph-structured data. At the same time, unsupervised learning of graph embeddings has benefited from the information contained in random walks. In this paper we propose a model, Network of GCNs (N-GCN), which marries these two lines of work. At its core, N-GCN trains multiple instances of GCNs over node pairs discovered at different distances in random walks, and learns a combination of the instance outputs which optimizes the classification objective. Our experiments show that our proposed N-GCN model achieves state-of-the-art performance on all of the challenging node classification tasks we consider: Cora, Citeseer, Pubmed, and PPI. In addition, our proposed method has other desirable properties, including generalization to recently proposed semi-supervised learning methods such as GraphSAGE, allowing us to propose N-SAGE, and resilience to adversarial input perturbations.", "target": ["Facciamo una rete di Graph Convolution Network, alimentando ciascuna con una potenza diversa della matrice di adiacenza, combinando tutta la loro rappresentazione in una sottorete di classificazione, raggiungendo lo stato dell'arte sulla classificazione semi-supervisionata dei nodi.", "Propone una nuova rete di GCN con due approcci: un layer completamente connesso sopra le feature impilate e un meccanismo di attention che usa un peso scalare per GCN.", "Presenta una rete di GCN che usa la statistica del cammino casuale per estrarre informazioni dai neighbor vicini e lontani nel grafo"]} +{"source": "Recent DNN pruning algorithms have succeeded in reducing the number of parameters in fully connected layers often with little or no drop in classification accuracy. However most of the existing pruning schemes either have to be applied during training or require a costly retraining procedure after pruning to regain classification accuracy. In this paper we propose a cheap pruning algorithm based on difference of convex (DC) optimisation. We also provide theoretical analysis for the growth in the Generalisation Error (GE) of the new pruned network. Our method can be used with any convex regulariser and allows for a controlled degradation in classification accuracy while being orders of magnitude faster than competing approaches. Experiments on common feedforward neural networks show that for sparsity levels above 90% our method achieves 10% higher classification accuracy compared to Hard Thresholding.", "target": ["Un algoritmo di pruning veloce per layer DNN fully connected con analisi teorica della degradazione dell'errore di generalizzazione.", "Presenta un algoritmo di pruning economico per layer densi di DNN.", "Propone una soluzione al problema del pruning delle DNN ponendo la funzione obiettivo Net-trim come una funzione Difference of convex (DC)."]} +{"source": "Action segmentation as a milestone towards building automatic systems to understand untrimmed videos has received considerable attention in the recent years. It is typically being modeled as a sequence labeling problem but contains intrinsic and sufficient differences than text parsing or speech processing. In this paper, we introduce a novel hybrid temporal convolutional and recurrent network (TricorNet), which has an encoder-decoder architecture: the encoder consists of a hierarchy of temporal convolutional kernels that capture the local motion changes of different actions; the decoder is a hierarchy of recurrent neural networks that are able to learn and memorize long-term action dependencies after the encoding stage. Our model is simple but extremely effective in terms of video sequence labeling. The experimental results on three public action segmentation datasets have shown that the proposed model achieves superior performance over the state of the art.", "target": ["Proponiamo una nuova rete temporale ibrida che raggiunge prestazioni allo stato dell'arte nella segmentazione di azioni video su tre dataset pubblici.", "Discute il problema della segmentazione delle azioni in video lunghi, fino a 10 minuti, utilizzando un'architettura di encoder-decoder convoluzionale temporale", "Propone una combinazione di rete temporale convoluzionale e ricorrente per la segmentazione di azioni video"]} +{"source": "Convolutional Neural Networks (CNNs) become deeper and deeper in recent years, making the study of model acceleration imperative. It is a common practice to employ a shallow network, called student, to learn from a deep one, which is termed as teacher. Prior work made many attempts to transfer different types of knowledge from teacher to student, however, there are two problems remaining unsolved. Firstly, the knowledge used by existing methods is highly dependent on task and dataset, limiting their applications. Secondly, there lacks an effective training scheme for the transfer process, leading to degradation of performance. In this work, we argue that feature is the most important knowledge from teacher. It is sufficient for student to just learn good features regardless of the target task. From this discovery, we further present an efficient learning strategy to mimic features stage by stage. Extensive experiments demonstrate the importance of features and show that the proposed approach significantly narrows down the gap between student and teacher, outperforming the state-of-the-art methods.", "target": ["Questo articolo propone di trasferire la conoscenza dal deep model a quello shallow imitando le feature fase per fase.", "Spiega un metodo di knowledge transfer fase per fase utilizzando diverse strutture di reti", "Questo articolo propone di dividere una rete in più parti e distillare ogni parte in modo sequenziale per migliorare le prestazioni di distillazione nelle deep teacher network"]} +{"source": "We augment adversarial training (AT) with worst case adversarial training (WCAT) which improves adversarial robustness by 11% over the current state- of-the-art result in the `2-norm on CIFAR-10. We interpret adversarial training as Total Variation Regularization, which is a fundamental tool in mathematical im- age processing, and WCAT as Lipschitz regularization, which appears in Image Inpainting. We obtain verifiable worst and average case robustness guarantees, based on the expected and maximum values of the norm of the gradient of the loss.", "target": ["Miglioramenti all'adversarial robustness, così come garanzie di robustezza dimostrabili, sono ottenuti aumentando l'adversarial training con una regolarizzazione Lipschitz trattabile", "Esplora l'arricchimento della loss di training con un termine aggiuntivo di regolarizzazione del gradiente per migliorare la robustezza dei modelli contro gli adversarial example", "Utilizza un trucco per semplificare l'adversarial loss con una loss in cui l'adversarial perturbation appare in forma chiusa."]} +{"source": "The task of Reading Comprehension with Multiple Choice Questions, requires a human (or machine) to read a given \\{\\textit{passage, question}\\} pair and select one of the $n$ given options. The current state of the art model for this task first computes a query-aware representation for the passage and then \\textit{selects} the option which has the maximum similarity with this representation. However, when humans perform this task they do not just focus on option selection but use a combination of \\textit{elimination} and \\textit{selection}. Specifically, a human would first try to eliminate the most irrelevant option and then read the document again in the light of this new information (and perhaps ignore portions corresponding to the eliminated option). This process could be repeated multiple times till the reader is finally ready to select the correct option. We propose \\textit{ElimiNet}, a neural network based model which tries to mimic this process. Specifically, it has gates which decide whether an option can be eliminated given the \\{\\textit{document, question}\\} pair and if so it tries to make the document representation orthogonal to this eliminatedd option (akin to ignoring portions of the document corresponding to the eliminated option). The model makes multiple rounds of partial elimination to refine the document representation and finally uses a selection module to pick the best option. We evaluate our model on the recently released large scale RACE dataset and show that it outperforms the current state of the art model on 7 out of the 13 question types in this dataset. Further we show that taking an ensemble of our \\textit{elimination-selection} based method with a \\textit{selection} based method gives us an improvement of 7\\% (relative) over the best reported performance on this dataset.", "target": ["Un modello che combina eliminazione e selezione per rispondere a domande a scelta multipla", "Dà un'elaborazione del Gated Attention Reader aggiungendo gate basati sull'eliminazione delle risposte nella multiple choice reading comprehension", "Questo articolo propone l'uso di un gate di eliminazione in architetture di modelli per task di comprensione della lettura, ma non raggiunge risultati allo stato dell'arte", "Questo articolo sostiene un nuovo modello di multi-choice reading comprehension basato sull'idea che alcune opzioni dovrebbero essere eliminate per dedurre una migliore rappresentazione del passaggio o della domanda."]} +{"source": "Humans are capable of attributing latent mental contents such as beliefs, or intentions to others. The social skill is critical in everyday life to reason about the potential consequences of their behaviors so as to plan ahead. It is known that humans use this reasoning ability recursively, i.e. considering what others believe about their own beliefs. In this paper, we start from level-$1$ recursion and introduce a probabilistic recursive reasoning (PR2) framework for multi-agent reinforcement learning. Our hypothesis is that it is beneficial for each agent to account for how the opponents would react to its future behaviors. Under the PR2 framework, we adopt variational Bayes methods to approximate the opponents' conditional policy, to which each agent finds the best response and then improve their own policy. We develop decentralized-training-decentralized-execution algorithms, PR2-Q and PR2-Actor-Critic, that are proved to converge in the self-play scenario when there is one Nash equilibrium. Our methods are tested on both the matrix game and the differential game, which have a non-trivial equilibrium where common gradient-based methods fail to converge. Our experiments show that it is critical to reason about how the opponents believe about what the agent believes. We expect our work to contribute a new idea of modeling the opponents to the multi-agent reinforcement learning community.", "target": ["Abbiamo proposto un nuovo framework di ragionamento ricorsivo probabilistico (PR2) per task di deep reinforcement learning multi-agente.", "Propone un nuovo approccio per il training completamente decentralizzato nel reinforcement learning multi-agente", "Il paper affronta il problema di dotare gli agenti RL di capacità di ragionamento ricorsivo in un ambiente multi-agente basato sull'ipotesi che il ragionamento ricorsivo sia vantaggioso per loro per convergere verso equilibri non banali", "L'articolo introduce un metodo di training decentralizzato per reinforcement learning multi-agente, dove gli agenti deducono le policy di altri agenti e usano i modelli dedotti per prendere decisioni."]} +{"source": "Due to the substantial computational cost, training state-of-the-art deep neural networks for large-scale datasets often requires distributed training using multiple computation workers. However, by nature, workers need to frequently communicate gradients, causing severe bottlenecks, especially on lower bandwidth connections. A few methods have been proposed to compress gradient for efficient communication, but they either suffer a low compression ratio or significantly harm the resulting model accuracy, particularly when applied to convolutional neural networks. To address these issues, we propose a method to reduce the communication overhead of distributed deep learning. Our key observation is that gradient updates can be delayed until an unambiguous (high amplitude, low variance) gradient has been calculated. We also present an efficient algorithm to compute the variance and prove that it can be obtained with negligible additional cost. We experimentally show that our method can achieve very high compression ratio while maintaining the result model accuracy. We also analyze the efficiency using computation and communication cost models and provide the evidence that this method enables distributed deep learning for many scenarios with commodity environments.", "target": ["Un nuovo algoritmo per ridurre l'overhead di comunicazione del deep learning distribuito distinguendo i gradienti \"non ambigui\".", "Propone un metodo di compressione del gradiente basato sulla varianza per ridurre l'overhead di comunicazione del deep learning distribuito", "Propone un nuovo modo di comprimere gli aggiornamenti del gradiente per SGD distribuito al fine di accelerare l'esecuzione complessiva", "Introduce un metodo di compressione del gradiente basato sulla varianza per un efficiente training distribuito delle reti neurali e per misurare l'ambiguità."]} +{"source": "In this work, we face the problem of unsupervised domain adaptation with a novel deep learning approach which leverages our finding that entropy minimization is induced by the optimal alignment of second order statistics between source and target domains. We formally demonstrate this hypothesis and, aiming at achieving an optimal alignment in practical cases, we adopt a more principled strategy which, differently from the current Euclidean approaches, deploys alignment along geodesics. Our pipeline can be implemented by adding to the standard classification loss (on the labeled source domain), a source-to-target regularizer that is weighted in an unsupervised and data-driven fashion. We provide extensive experiments to assess the superiority of our framework on standard domain and modality adaptation benchmarks.", "target": ["Una nuova tecnica non supervisionata di deep domain adaptation che unifica in modo efficiente l'allineamento delle correlazioni e la minimizzazione dell'entropia", "Migliora l'approccio di allineamento delle correlazioni al domain adaptation sostituendo la distanza euclidea con la distanza log-euclidea geodetica tra due matrici di covarianza, e selezionando automaticamente il costo di bilanciamento in base all'entropia sul dominio di destinazione.", "Proposta per l'allineamento di correlazione a minima entropia, un algoritmo di domain adaptation unsupervised che collega insieme la minimizzazione dell'entropia e i metodi di allineamento della correlazione."]} +{"source": "Catastrophic interference has been a major roadblock in the research of continual learning. Here we propose a variant of the back-propagation algorithm, \"Conceptor-Aided Backprop\" (CAB), in which gradients are shielded by conceptors against degradation of previously learned tasks. Conceptors have their origin in reservoir computing, where they have been previously shown to overcome catastrophic forgetting. CAB extends these results to deep feedforward networks. On the disjoint and permuted MNIST tasks, CAB outperforms two other methods for coping with catastrophic interference that have recently been proposed.", "target": ["Proponiamo una variante dell'algoritmo di backpropagation, in cui i gradienti sono schermati dai conceptor per prevenire la degradazione dei task precedentemente appresi.", "Questo articolo applica la nozione di conceptor, una forma di regolarizzatore, per prevenire il forgetting nel continual learning, in particolare nel training di reti neurali su task sequenziali.", "Introduce un metodo per imparare nuovi task, senza interferire con i task precedenti, usando i conceptor."]} +{"source": "Recent advances in neural Sequence-to-Sequence (Seq2Seq) models reveal a purely data-driven approach to the response generation task. Despite its diverse variants and applications, the existing Seq2Seq models are prone to producing short and generic replies, which blocks such neural network architectures from being utilized in practical open-domain response generation tasks. In this research, we analyze this critical issue from the perspective of the optimization goal of models and the specific characteristics of human-to-human conversational corpora. Our analysis is conducted by decomposing the goal of Neural Response Generation (NRG) into the optimizations of word selection and ordering. It can be derived from the decomposing that Seq2Seq based NRG models naturally tend to select common words to compose responses, and ignore the semantic of queries in word ordering. On the basis of the analysis, we propose a max-marginal ranking regularization term to avoid Seq2Seq models from producing the generic and uninformative responses. The empirical experiments on benchmarks with several metrics have validated our analysis and proposed methodology.", "target": ["Analizzare la ragione per cui i modelli neurali che generano risposte preferiscono risposte universali; proponiamo un metodo per evitarlo.", "Indaga il problema delle risposte universali che affligge i modelli di generazione neurale Seq2Seq", "L'articolo cerca di migliorare il task di generazione di risposte neurali de-enfatizzando le risposte comuni utilizzando la modifica della loss e presentando le risposte comuni/universali durante la fase di training."]} +{"source": "The ability to generate natural language sequences from source code snippets has a variety of applications such as code summarization, documentation, and retrieval. Sequence-to-sequence (seq2seq) models, adopted from neural machine translation (NMT), have achieved state-of-the-art performance on these tasks by treating source code as a sequence of tokens. We present code2seq: an alternative approach that leverages the syntactic structure of programming languages to better encode source code. Our model represents a code snippet as the set of compositional paths in its abstract syntax tree (AST) and uses attention to select the relevant paths while decoding. We demonstrate the effectiveness of our approach for two tasks, two programming languages, and four datasets of up to 16M examples. Our model significantly outperforms previous models that were specifically designed for programming languages, as well as general state-of-the-art NMT models. An interactive online demo of our model is available at http://code2seq.org. Our code, data and trained models are available at http://github.com/tech-srl/code2seq.", "target": ["Sfruttiamo la struttura sintattica del codice sorgente per generare sequenze di linguaggio naturale.", "Presenta un metodo per generare sequenze dal codice analizzando e producendo un albero di sintassi", "Questo articolo introduce una codifica basata su AST per il codice di programmazione e mostra la sua efficacia nei task di extreme code summarization e code captioning.", "Questo articolo presenta un nuovo modello code-to-sequence che sfrutta la struttura sintattica dei linguaggi di programmazione per codificare frammenti di codice sorgente e poi decodificarli in linguaggio naturale"]} +{"source": "We propose a novel attention mechanism to enhance Convolutional Neural Networks for fine-grained recognition. The proposed mechanism reuses CNN feature activations to find the most informative parts of the image at different depths with the help of gating mechanisms and without part annotations. Thus, it can be used to augment any layer of a CNN to extract low- and high-level local information to be more discriminative. Differently, from other approaches, the mechanism we propose just needs a single pass through the input and it can be trained end-to-end through SGD. As a consequence, the proposed mechanism is modular, architecture-independent, easy to implement, and faster than iterative approaches. Experiments show that, when augmented with our approach, Wide Residual Networks systematically achieve superior performance on each of five different fine-grained recognition datasets: the Adience age and gender recognition benchmark, Caltech-UCSD Birds-200-2011, Stanford Dogs, Stanford Cars, and UEC Food-100, obtaining competitive and state-of-the-art scores.", "target": ["Miglioriamo le CNN con un nuovo meccanismo di attention per il riconoscimento preciso. Ciò risulta in prestazioni superiori su 5 dataset.", "Descrive un nuovo meccanismo attenzionale applicato al riconoscimento preciso che migliora costantemente l'accuratezza del riconoscimento della baseline", "Questo articolo propone un meccanismo di attention feed-forward per la classificazione fine-grained delle immagini ", "Questo articolo presenta un interessante meccanismo di attention per la classificazione fine-grained delle immagini."]} +{"source": "Most existing GANs architectures that generate images use transposed convolution or resize-convolution as their upsampling algorithm from lower to higher resolution feature maps in the generator. We argue that this kind of fixed operation is problematic for GANs to model objects that have very different visual appearances. We propose a novel adaptive convolution method that learns the upsampling algorithm based on the local context at each location to address this problem. We modify a baseline GANs architecture by replacing normal convolutions with adaptive convolutions in the generator. Experiments on CIFAR-10 dataset show that our modified models improve the baseline model by a large margin. Furthermore, our models achieve state-of-the-art performance on CIFAR-10 and STL-10 datasets in the unsupervised setting.", "target": ["Sostituiamo le normali convoluzioni con le convoluzioni adattive per migliorare il generatore delle GAN.", "Propone di sostituire le convoluzioni nel generatore con un blocco di convoluzione adattiva che impara a generare pesi e bias di convoluzione delle operazioni di upsampling in modo adattivo rispetto alla posizione del pixel", "Utilizza Adaptive Convolution nel contesto delle GAN con un blocco chiamato AdaConvBlock che sostituisce la normale Convoluzione, questo dà più contesto locale per kernel weight in modo da poter generare oggetti localmente flessibili."]} +{"source": "Techniques such as ensembling and distillation promise model quality improvements when paired with almost any base model. However, due to increased test-time cost (for ensembles) and increased complexity of the training pipeline (for distillation), these techniques are challenging to use in industrial settings. In this paper we explore a variant of distillation which is relatively straightforward to use as it does not require a complicated multi-stage setup or many new hyperparameters. Our first claim is that online distillation enables us to use extra parallelism to fit very large datasets about twice as fast. Crucially, we can still speed up training even after we have already reached the point at which additional parallelism provides no benefit for synchronous or asynchronous stochastic gradient descent. Two neural networks trained on disjoint subsets of the data can share knowledge by encouraging each model to agree with the predictions the other model would have made. These predictions can come from a stale version of the other model so they can be safely computed using weights that only rarely get transmitted. Our second claim is that online distillation is a cost-effective way to make the exact predictions of a model dramatically more reproducible. We support our claims using experiments on the Criteo Display Ad Challenge dataset, ImageNet, and the largest to-date dataset used for neural language modeling, containing $6\\times 10^{11}$ tokens and based on the Common Crawl repository of web data.", "target": ["Eseguiamo esperimenti su larga scala per dimostrare che una semplice variante online di distillation può aiutarci a scalare il training di reti neurali distribuite a più macchine.", "Propone un metodo per scalare il training distribuito oltre i limiti attuali della mini-batch stochastic gradient descent", "Proposta di un metodo di distillation online chiamato co-distillation, applicato su scala, in cui due diversi modelli sono addestrati per avvicinare le proprie predizioni a quelle dell'altro modello oltre a minimizzare la propria loss.", "Viene introdotta una tecnica di distillation online per accelerare gli algoritmi tradizionali per il training di reti neurali distribuite su larga scala"]} +{"source": "Support Vector Machines (SVMs) are one of the most popular algorithms for classification and regression analysis. Despite their popularity, even efficient implementations have proven to be computationally expensive to train at a large-scale, especially in streaming settings. In this paper, we propose a novel coreset construction algorithm for efficiently generating compact representations of massive data sets to speed up SVM training. A coreset is a weighted subset of the original data points such that SVMs trained on the coreset are provably competitive with those trained on the original (massive) data set. We provide both lower and upper bounds on the number of samples required to obtain accurate approximations to the SVM problem as a function of the complexity of the input data. Our analysis also establishes sufficient conditions on the existence of sufficiently compact and representative coresets for the SVM problem. We empirically evaluate the practical effectiveness of our algorithm against synthetic and real-world data sets.", "target": ["Presentiamo un algoritmo per accelerare il training SVM su massicci dataset costruendo rappresentazioni compatte che forniscono un'inferenza efficiente e approssimata in modo dimostrato.", "Studia l'approccio del coreset per SVM e mira a campionare un piccolo insieme di punti ponderati tali che la loss valutata sui quei punti approssimi in modo dimostrabile quella sull'intero dataset", "L'articolo suggerisce una costruzione di Coreset basata sull'importance sampling per rappresentare grandi moli di dati di training per SVM"]} +{"source": "The sign stochastic gradient descent method (signSGD) utilizes only the sign of the stochastic gradient in its updates. Since signSGD carries out one-bit quantization of the gradients, it is extremely practical for distributed optimization where gradients need to be aggregated from different processors. For the first time, we establish convergence rates for signSGD on general non-convex functions under transparent conditions. We show that the rate of signSGD to reach first-order critical points matches that of SGD in terms of number of stochastic gradient calls, up to roughly a linear factor in the dimension. We carry out simple experiments to explore the behaviour of sign gradient descent (without the stochasticity) close to saddle points and show that it often helps completely avoid them without using either stochasticity or curvature information.", "target": ["Dimostriamo un tasso di convergenza non convesso per il metodo del sign stochastic gradient. L'algoritmo ha collegamenti con algoritmi come Adam e Rprop, così come con schemi di quantizzazione del gradiente usati nell'apprendimento automatico distribuito.", "Fornito un'analisi di convergenza dell'algoritmo Sign SGD per i casi non covessivi", "L'articolo esplora un algoritmo che utilizza il segno dei gradienti invece dei gradienti effettivi per il training di deep model"]} +{"source": "Deep learning has found numerous applications thanks to its versatility and accuracy on pattern recognition problems such as visual object detection. Learning and inference in deep neural networks, however, are memory and compute intensive and so improving efficiency is one of the major challenges for frameworks such as PyTorch, Tensorflow, and Caffe. While the efficiency problem can be partially addressed with specialized hardware and its corresponding proprietary libraries, we believe that neural network acceleration should be transparent to the user and should support all hardware platforms and deep learning libraries. To this end, we introduce a transparent middleware layer for neural network acceleration. The system is built around a compiler for deep learning, allowing one to combine device-specific libraries and custom optimizations while supporting numerous hardware devices. In contrast to other projects, we explicitly target the optimization of both prediction and training of neural networks. We present the current development status and some preliminary but encouraging results: on a standard x86 server, using CPUs our system achieves a 11.8x speed-up for inference and a 8.0x for batched-prediction (128); on GPUs we achieve a 1.7x and 2.3x speed-up respectively.", "target": ["Introduciamo un middleware trasparente per l'accelerazione delle reti neurali, con un proprio motore di compilazione, raggiungendo fino a 11.8x di velocità sulle CPU e 2.3x sulle GPU.", "Questo articolo propone un livello di middleware trasparente per l'accelerazione delle reti neurali e ottiene alcuni risultati di accelerazione su architetture CPU e GPU di base"]} +{"source": "Performance of neural networks can be significantly improved by encoding known invariance for particular tasks. Many image classification tasks, such as those related to cellular imaging, exhibit invariance to rotation. In particular, to aid convolutional neural networks in learning rotation invariance, we consider a simple, efficient conic convolutional scheme that encodes rotational equivariance, along with a method for integrating the magnitude response of the 2D-discrete-Fourier transform (2D-DFT) to encode global rotational invariance. We call our new method the Conic Convolution and DFT Network (CFNet). We evaluated the efficacy of CFNet as compared to a standard CNN and group-equivariant CNN (G-CNN) for several different image classification tasks and demonstrated improved performance, including classification accuracy, computational efficiency, and its robustness to hyperparameter selection. Taken together, we believe CFNet represents a new scheme that has the potential to improve many imaging analysis applications.", "target": ["Proponiamo la convoluzione conica e la 2D-DFT per codificare l'equivarianza di rotazione in una rete neurale.", "Nel contesto della classificazione delle immagini, l'articolo propone un'architettura di rete neurale convoluzionale di feature map equivarianti alla rotazione che sono infine rese invarianti alla rotazione usando il modulo della trasformata discreta di Fourier 2D (DFT).", "Gli autori forniscono una rete neurale invariante alla rotazione attraverso la combinazione di convoluzione conica e 2D-DFT"]} +{"source": "The problem of visual metamerism is defined as finding a family of perceptually indistinguishable, yet physically different images. In this paper, we propose our NeuroFovea metamer model, a foveated generative model that is based on a mixture of peripheral representations and style transfer forward-pass algorithms. Our gradient-descent free model is parametrized by a foveated VGG19 encoder-decoder which allows us to encode images in high dimensional space and interpolate between the content and texture information with adaptive instance normalization anywhere in the visual field. Our contributions include: 1) A framework for computing metamers that resembles a noisy communication system via a foveated feed-forward encoder-decoder network – We observe that metamerism arises as a byproduct of noisy perturbations that partially lie in the perceptual null space; 2) A perceptual optimization scheme as a solution to the hyperparametric nature of our metamer model that requires tuning of the image-texture tradeoff coefficients everywhere in the visual field which are a consequence of internal noise; 3) An ABX psychophysical evaluation of our metamers where we also find that the rate of growth of the receptive fields in our model match V1 for reference metamers and V2 between synthesized samples. Our model also renders metamers at roughly a second, presenting a ×1000 speed-up compared to the previous work, which now allows for tractable data-driven metamer experiments.", "target": ["Introduciamo un nuovo framework feed-forward per generare metameri visivi", "Propone un modello NeuroFovea per la generazione di metameri point-of-fixation utilizzando un approccio di transfer di stile tramite un'architettura Encoder-Decoder", "Un'analisi del metamerismo e un modello capace di produrre rapidamente metameri di valore per la psicofisica sperimentale e altri domini.", "L'articolo propone un metodo veloce per generare metameri visivi - immagini fisicamente diverse che non possono essere distinte da un originale - attraverso un transfer di stile foveated, veloce e arbitrario"]} +{"source": "Past works have shown that, somewhat surprisingly, over-parametrization can help generalization in neural networks. Towards explaining this phenomenon, we adopt a margin-based perspective. We establish: 1) for multi-layer feedforward relu networks, the global minimizer of a weakly-regularized cross-entropy loss has the maximum normalized margin among all networks, 2) as a result, increasing the over-parametrization improves the normalized margin and generalization error bounds for deep networks. In the case of two-layer networks, an infinite-width neural network enjoys the best generalization guarantees. The typical infinite feature methods are kernel methods; we compare the neural net margin with that of kernel methods and construct natural instances where kernel methods have much weaker generalization guarantees. We validate this gap between the two approaches empirically. Finally, this infinite-neuron viewpoint is also fruitful for analyzing optimization. We show that a perturbed gradient flow on infinite-size networks finds a global optimizer in polynomial time.", "target": ["Mostriamo che il training delle reti relu feedforward con un regolarizzatore debole risulta in un margine massimo e analizziamo le implicazioni di questo risultato.", "Il paper studia la teoria del margine per gli insiemi neurali e mostra che il margine massimo è monotonicamente crescente nella dimensione della rete", "Questo articolo studia il bias implicito dei minimizzatori di una cross-entropy loss regolarizzata di una rete a due layer con attivazioni ReLU, ottenendo un limite superiore di generalizzazione che non aumenta con la dimensione della rete."]} +{"source": "We propose a distributed architecture for deep reinforcement learning at scale, that enables agents to learn effectively from orders of magnitude more data than previously possible. The algorithm decouples acting from learning: the actors interact with their own instances of the environment by selecting actions according to a shared neural network, and accumulate the resulting experience in a shared experience replay memory; the learner replays samples of experience and updates the neural network. The architecture relies on prioritized experience replay to focus only on the most significant data generated by the actors. Our architecture substantially improves the state of the art on the Arcade Learning Environment, achieving better final performance in a fraction of the wall-clock training time.", "target": ["Un'architettura distribuita per reinforcement learning profondo su scala, utilizzando la generazione di dati in parallelo per migliorare lo stato dell'arte sul benchmark Arcade Learning Environment in una frazione del tempo di training in wall-clock degli approcci precedenti.", "Esamina un sistema di deep RL distribuito in cui le esperienze, piuttosto che i gradienti, sono condivisi tra i lavori paralleli e il learner centralizzato", "Un approccio parallelo al training di DQN, basato sull'idea di avere più actor che raccolgono dati in parallelo mentre un singolo learner addestra il modello da esperienze campionate dalla memoria centrale di replay.", "Questo articolo propone un'architettura distribuita per deep reinforcement learning su scala, concentrandosi sull'aggiunta di parallelizzazione nell'algoritmo dell'actor nel framework Prioritized Experience Replay"]} +{"source": "Designing neural networks for continuous-time stochastic processes is challenging, especially when observations are made irregularly. In this article, we analyze neural networks from a frame theoretic perspective to identify the sufficient conditions that enable smoothly recoverable representations of signals in L^2(R). Moreover, we show that, under certain assumptions, these properties hold even when signals are irregularly observed. As we converge to the family of (convolutional) neural networks that satisfy these conditions, we show that we can optimize our convolution filters while constraining them so that they effectively compute a Discrete Wavelet Transform. Such a neural network can efficiently divide the time-axis of a signal into orthogonal sub-spaces of different temporal scale and localization. We evaluate the resulting neural network on an assortment of synthetic and real-world tasks: parsimonious auto-encoding, video classification, and financial forecasting.", "target": ["Architetture neurali che forniscono rappresentazioni di segnali osservati in modo irregolare e che permettono la ricostruzione del segnale.", "Dimostra che le reti neurali convoluzionali con funzione di attivazione Leaky ReLU sono strutture non lineari, con risultati simili per serie temporali campionate in modo non uniforme", "Questo articolo considera le reti neurali su serie temporali e mostra che i primi filtri convoluzionali possono essere scelti per rappresentare una trasformata wavelet discreta."]} +{"source": "Most state-of-the-art neural machine translation systems, despite being different in architectural skeletons (e.g., recurrence, convolutional), share an indispensable feature: the Attention. However, most existing attention methods are token-based and ignore the importance of phrasal alignments, the key ingredient for the success of phrase-based statistical machine translation. In this paper, we propose novel phrase-based attention methods to model n-grams of tokens as attention entities. We incorporate our phrase-based attentions into the recently proposed Transformer network, and demonstrate that our approach yields improvements of 1.3 BLEU for English-to-German and 0.5 BLEU for German-to-English translation tasks, and 1.75 and 1.35 BLEU points in English-to-Russian and Russian-to-English translation tasks on WMT newstest2014 using WMT’16 training data.", "target": ["Meccanismi di attention basati sulla frase per assegnare l'attention sulle frasi, ottenendo allineamenti di attention token-to-phrase, phrase-to-token, phrase-to-phrase, oltre alle attention token-token esistenti.", "L'articolo presenta un meccanismo di attention che calcola una somma pesata non solo su singoli token ma su ngrammi (frasi)."]} +{"source": "Intuitively, unfamiliarity should lead to lack of confidence. In reality, current algorithms often make highly confident yet wrong predictions when faced with unexpected test samples from an unknown distribution different from training. Unlike domain adaptation methods, we cannot gather an \"unexpected dataset\" prior to test, and unlike novelty detection methods, a best-effort original task prediction is still expected. We compare a number of methods from related fields such as calibration and epistemic uncertainty modeling, as well as two proposed methods that reduce overconfident errors of samples from an unknown novel distribution without drastically increasing evaluation time: (1) G-distillation, training an ensemble of classifiers and then distill into a single model using both labeled and unlabeled examples, or (2) NCR, reducing prediction confidence based on its novelty detection score. Experimentally, we investigate the overconfidence problem and evaluate our solution by creating \"familiar\" and \"novel\" test splits, where \"familiar\" are identically distributed with training and \"novel\" are not. We discover that calibrating using temperature scaling on familiar data is the best single-model method for improving novel confidence, followed by our proposed methods. In addition, some methods' NLL performance are roughly equivalent to a regularly trained model with certain degree of smoothing. Calibrating can also reduce confident errors, for example, in gender recognition by 95% on demographic groups different from the training data.", "target": ["Le reti profonde hanno maggiori probabilità di sbagliare in modo consistente quando la fase di testing contiene dati inaspettati. Proponiamo una metodologia sperimentale per studiare il problema, e due metodi per ridurre gli errori aventi alta confidenza su distribuzioni di input sconosciute.", "Propone due idee per ridurre le predizioni errate troppo sicure: \"G-distillation\" di un insieme con dati extra non supervisionati e Novelty Confidence Reduction con un novelty detector", "Gli autori propongono due metodi per stimare la fiducia di classificazione su nuove distribuzioni di dati non visti. La prima idea è quella di utilizzare metodi di ensemble come approccio di base per aiutare a identificare i casi incerti e poi utilizzare metodi di distillation per ridurre l'ensemble in un singolo modello che imita il comportamento dell'ensemble. La seconda idea è di usare un classificatore novelty detector e pesare l'output della rete in base al punteggio di novelty."]} +{"source": "Progress in deep learning is slowed by the days or weeks it takes to train large models. The natural solution of using more hardware is limited by diminishing returns, and leads to inefficient use of additional resources. In this paper, we present a large batch, stochastic optimization algorithm that is both faster than widely used algorithms for fixed amounts of computation, and also scales up substantially better as more computational resources become available. Our algorithm implicitly computes the inverse Hessian of each mini-batch to produce descent directions; we do so without either an explicit approximation to the Hessian or Hessian-vector products. We demonstrate the effectiveness of our algorithm by successfully training large ImageNet models (InceptionV3, ResnetV1-50, ResnetV1-101 and InceptionResnetV2) with mini-batch sizes of up to 32000 with no loss in validation error relative to current baselines, and no increase in the total number of steps. At smaller mini-batch sizes, our optimizer improves the validation error in these models by 0.8-0.9\\%. Alternatively, we can trade off this accuracy to reduce the number of training steps needed by roughly 10-30\\%. Our work is practical and easily usable by others -- only one hyperparameter (learning rate) needs tuning, and furthermore, the algorithm is as computationally cheap as the commonly used Adam optimizer.", "target": ["Descriviamo un pratico algoritmo di ottimizzazione per deep neural network che funziona più velocemente e genera modelli migliori rispetto agli algoritmi ampiamente utilizzati.", "Il paper propone un nuovo algoritmo in cui si afferma di usare implicitamente l'Hessiana e si usa una motivazione dalle serie di potenza", "Presenta un nuovo algoritmo del 2° ordine che usa implicitamente l'informazione sulla curvatura e mostra l'intuizione dietro gli schemi di approssimazione negli algoritmi e convalida l'euristica in vari esperimenti."]} +{"source": "Recent work has shown that performing inference with fast, very-low-bitwidth (e.g., 1 to 2 bits) representations of values in models can yield surprisingly accurate results. However, although 2-bit approximated networks have been shown to be quite accurate, 1 bit approximations, which are twice as fast, have restrictively low accuracy. We propose a method to train models whose weights are a mixture of bitwidths, that allows us to more finely tune the accuracy/speed trade-off. We present the “middle-out” criterion for determining the bitwidth for each value, and show how to integrate it into training models with a desired mixture of bitwidths. We evaluate several architectures and binarization techniques on the ImageNet dataset. We show that our heterogeneous bitwidth approximation achieves superlinear scaling of accuracy with bitwidth. Using an average of only 1.4 bits, we are able to outperform state-of-the-art 2-bit architectures.", "target": ["Introduciamo l'approssimazione della larghezza di bit frazionaria e mostriamo che ha dei vantaggi significativi.", "Suggerisce un metodo per variare il grado di quantizzazione in una rete neurale durante la forward phase", "Mantenere la precisione di una network a 2 bit mentre si usano meno di 2 bit per i pesi"]} +{"source": "Pruning units in a deep network can help speed up inference and training as well as reduce the size of the model. We show that bias propagation is a pruning technique which consistently outperforms the common approach of merely removing units, regardless of the architecture and the dataset. We also show how a simple adaptation to an existing scoring function allows us to select the best units to prune. Finally, we show that the units selected by the best performing scoring functions are somewhat consistent over the course of training, implying the dead parts of the network appear during the stages of training.", "target": ["La sostituzione media è un metodo efficiente per migliorare la loss dopo il pruning e le funzioni di punteggio basate sull'approssimazione di Taylor funzionano meglio con i valori assoluti.", "Propone un semplice miglioramento dei metodi per il pruning delle unità utilizzando la \"sostituzione media\".", "Questo articolo presenta una strategia di pruning di mean-replacement e utilizza l'espansione di Taylor a valore assoluto come funzione di punteggio per il pruning"]} +{"source": "Due to the phenomenon of “posterior collapse,” current latent variable generative models pose a challenging design choice that either weakens the capacity of the decoder or requires altering the training objective. We develop an alternative that utilizes the most powerful generative models as decoders, optimize the variational lower bound, and ensures that the latent variables preserve and encode useful information. Our proposed δ-VAEs achieve this by constraining the variational family for the posterior to have a minimum distance to the prior. For sequential latent variable models, our approach resembles the classic representation learning approach of slow feature analysis. We demonstrate our method’s efficacy at modeling text on LM1B and modeling images: learning representations, improving sample quality, and achieving state of the art log-likelihood on CIFAR-10 and ImageNet 32 × 32.", "target": ["Evitare il posterior collapse limitando inferiormente il tasso.", "Presenta un approccio per prevenire il posterior collapse nei VAE limitando la famiglia dell'approssimazione variazionale al posterior", "Questo articolo introduce un vincolo sulla famiglia dei posterior variazionali in modo che il termine KL possa essere controllato per combattere il posterior collapse nei deep generative model come i VAE"]} +{"source": "Mini-batch gradient descent and its variants are commonly used in deep learning. The principle of mini-batch gradient descent is to use noisy gradient calculated on a batch to estimate the real gradient, thus balancing the computation cost per iteration and the uncertainty of noisy gradient. However, its batch size is a fixed hyper-parameter requiring manual setting before training the neural network. Yin et al. (2017) proposed a batch adaptive stochastic gradient descent (BA-SGD) that can dynamically choose a proper batch size as learning proceeds. We extend the BA-SGD to momentum algorithm and evaluate both the BA-SGD and the batch adaptive momentum (BA-Momentum) on two deep learning tasks from natural language processing to image classification. Experiments confirm that batch adaptive methods can achieve a lower loss compared with mini-batch methods after scanning the same epochs of data. Furthermore, our BA-Momentum is more robust against larger step sizes, in that it can dynamically enlarge the batch size to reduce the larger uncertainty brought by larger step sizes. We also identified an interesting phenomenon, batch size boom. The code implementing batch adaptive framework is now open source, applicable to any gradient-based optimization problems.", "target": ["Abbiamo sviluppato un momentum batch adaptive che può raggiungere una loss inferiore rispetto ai metodi mini-batch dopo la scansione delle stesse epoche di dati, ed è più robusto verso grandi step size.", "Questo articolo affronta il problema di sintonizzare automaticamente la batch size durante il training in deep learning, e sostiene di estendere l'SGD batch adaptive al momentum adattivo e di adottare gli algoritmi ai problemi delle reti neurali complesse.", "L'articolo propone di generalizzare un algoritmo che esegue SGD con adaptive batch size aggiungendo momentum alla utility function"]} +{"source": "Deep learning models for graphs have advanced the state of the art on many tasks. Despite their recent success, little is known about their robustness. We investigate training time attacks on graph neural networks for node classification that perturb the discrete graph structure. Our core principle is to use meta-gradients to solve the bilevel problem underlying training-time attacks, essentially treating the graph as a hyperparameter to optimize. Our experiments show that small graph perturbations consistently lead to a strong decrease in performance for graph convolutional networks, and even transfer to unsupervised embeddings. Remarkably, the perturbations created by our algorithm can misguide the graph neural networks such that they perform worse than a simple baseline that ignores all relational information. Our attacks do not assume any knowledge about or access to the target classifiers.", "target": ["Usiamo i meta-gradienti per attaccare la procedura di training delle deep neural network per i grafi.", "Studia il problema dell'apprendimento di un migliore parametro per un grafo poisoned che può massimizzare la loss di una graph neural network.", "Un algoritmo per alterare la struttura del grafo aggiungendo/eliminando archi in modo da degradare la performance globale della classificazione dei nodi, con l'idea di usare il meta-learning per risolvere il problema di ottimizzazione a due livelli."]} +{"source": "Numerous models for grounded language understanding have been recently proposed, including (i) generic models that can be easily adapted to any given task and (ii) intuitively appealing modular models that require background knowledge to be instantiated. We compare both types of models in how much they lend themselves to a particular form of systematic generalization. Using a synthetic VQA test, we evaluate which models are capable of reasoning about all possible object pairs after training on only a small subset of them. Our findings show that the generalization of modular models is much more systematic and that it is highly sensitive to the module layout, i.e. to how exactly the modules are connected. We furthermore investigate if modular models that generalize well could be made more end-to-end by learning their layout and parametrization. We find that end-to-end methods from prior work often learn inappropriate layouts or parametrizations that do not facilitate systematic generalization. Our results suggest that, in addition to modularity, systematic generalization in language understanding may require explicit regularizers or priors.", "target": ["Mostriamo che i modelli strutturati modulari sono i migliori in termini di generalizzazione sistematica e che le loro versioni end-to-end non generalizzano altrettanto bene.", "Questo articolo valuta la generalizzazione sistemica tra reti neurali modulari e modelli generici attraverso l'introduzione di un nuovo dataset di spatial reasoning ", "Una valutazione empirica mirata della generalizzazione nei modelli per il ragionamento visivo, focalizzata sul problema del riconoscimento delle triple (oggetto, relazione, oggetto) in scene sintetiche con lettere e numeri."]} +{"source": "The behavioral dynamics of multi-agent systems have a rich and orderly structure, which can be leveraged to understand these systems, and to improve how artificial agents learn to operate in them. Here we introduce Relational Forward Models (RFM) for multi-agent learning, networks that can learn to make accurate predictions of agents' future behavior in multi-agent environments. Because these models operate on the discrete entities and relations present in the environment, they produce interpretable intermediate representations which offer insights into what drives agents' behavior, and what events mediate the intensity and valence of social interactions. Furthermore, we show that embedding RFM modules inside agents results in faster learning systems compared to non-augmented baselines. As more and more of the autonomous systems we develop and interact with become multi-agent in nature, developing richer analysis tools for characterizing how and why agents make decisions is increasingly necessary. Moreover, developing artificial agents that quickly and safely learn to coordinate with one another, and with humans in shared environments, is crucial.", "target": ["I Relational Forward Model per l'apprendimento multi-agente fanno predizioni accurate del comportamento futuro degli agenti, producono rappresentazioni intepretabili e possono essere utilizzati all'interno degli agenti.", "Un modo per ridurre la varianza nell'apprendimento model-free avendo un modello esplicito, con un'architettura simile a graph convolutional network, delle azioni che gli altri agenti compiranno.", "Predire il comportamento multi-agente usando un modello relazionale forward con una componente ricorrente, superando due baseline e due ablazioni"]} +{"source": "We show that gradient descent on an unregularized logistic regression problem, for almost all separable datasets, converges to the same direction as the max-margin solution. The result generalizes also to other monotone decreasing loss functions with an infimum at infinity, and we also discuss a multi-class generalizations to the cross entropy loss. Furthermore, we show this convergence is very slow, and only logarithmic in the convergence of the loss itself. This can help explain the benefit of continuing to optimize the logistic or cross-entropy loss even after the training error is zero and the training loss is extremely small, and, as we show, even if the validation loss increases. Our methodology can also aid in understanding implicit regularization in more complex models and with other optimization methods.", "target": ["La soluzione normalizzata della gradient descent sulla regressione logistica (o una loss che decade in modo simile) converge lentamente alla soluzione L2 max margin su dati separabili.", "L'articolo offre una prova formale che la gradient descent sulla loss logistica converge molto lentamente alla soluzione SVM hard nel caso in cui i dati siano linearmente separabili.", "Questo articolo si concentra sulla caratterizzazione del comportamento della minimizzazione della log-loss su dati linearmente separabili, e mostra che la log-loss, minimizzata con la discesa del gradiente, porta a convergere alla soluzione di max-margin."]} +{"source": "Despite impressive performance as evaluated on i.i.d. holdout data, deep neural networks depend heavily on superficial statistics of the training data and are liable to break under distribution shift. For example, subtle changes to the background or texture of an image can break a seemingly powerful classifier. Building on previous work on domain generalization, we hope to produce a classifier that will generalize to previously unseen domains, even when domain identifiers are not available during training. This setting is challenging because the model may extract many distribution-specific (superficial) signals together with distribution-agnostic (semantic) signals. To overcome this challenge, we incorporate the gray-level co-occurrence matrix (GLCM) to extract patterns that our prior knowledge suggests are superficial: they are sensitive to the texture but unable to capture the gestalt of an image. Then we introduce two techniques for improving our networks' out-of-sample performance. The first method is built on the reverse gradient method that pushes our model to learn representations from which the GLCM representation is not predictable. The second method is built on the independence introduced by projecting the model's representation onto the subspace orthogonal to GLCM representation's. We test our method on the battery of standard domain generalization data sets and, interestingly, achieve comparable or better performance as compared to other domain generalization methods that explicitly require samples from the target distribution for training.", "target": ["Basandoci sul lavoro precedente sulla generalizzazione del dominio, speriamo di produrre un classificatore che generalizzi a domini precedentemente non visti, anche quando gli identificatori di dominio non sono disponibili durante il training.", "Un approccio di generalizzazione del dominio per rivelare informazioni semantiche basate su uno schema di proiezione lineare da CNN e layer di uscita NGLCM.", "L'articolo propone un approccio unsupervised per identificare le feature delle immagini che non sono significative per i task di classificazione delle immagini"]} +{"source": "In this paper, we conduct an intriguing experimental study about the physical adversarial attack on object detectors in the wild. In particular, we learn a camouflage pattern to hide vehicles from being detected by state-of-the-art convolutional neural network based detectors. Our approach alternates between two threads. In the first, we train a neural approximation function to imitate how a simulator applies a camouflage to vehicles and how a vehicle detector performs given images of the camouflaged vehicles. In the second, we minimize the approximated detection score by searching for the optimal camouflage. Experiments show that the learned camouflage can not only hide a vehicle from the image-based detectors under many test cases but also generalizes to different environments, vehicles, and object detectors.", "target": ["Proponiamo un metodo per imparare la mimetizzazione fisica del veicolo per attaccare in modo adversarial gli object detector in natura. Troviamo il nostro camuffamento efficace e trasferibile.", "Gli autori studiano il problema dell'apprendimento di un modello di mimetizzazione che, quando applicato a un veicolo simulato, impedirà a un rilevatore di oggetti di rilevarlo.", "Questo articolo mira all'adversarial learning per interferire nel rilevamento di auto apprendendo modelli di mimetizzazione"]} +{"source": "As deep learning-based classifiers are increasingly adopted in real-world applications, the importance of understanding how a particular label is chosen grows. Single decision trees are an example of a simple, interpretable classifier, but are unsuitable for use with complex, high-dimensional data. On the other hand, the variational autoencoder (VAE) is designed to learn a factored, low-dimensional representation of data, but typically encodes high-likelihood data in an intrinsically non-separable way. We introduce the differentiable decision tree (DDT) as a modular component of deep networks and a simple, differentiable loss function that allows for end-to-end optimization of a deep network to compress high-dimensional data for classification by a single decision tree. We also explore the power of labeled data in a supervised VAE (SVAE) with a Gaussian mixture prior, which leverages label information to produce a high-quality generative model with improved bounds on log-likelihood. We combine the SVAE with the DDT to get our classifier+VAE (C+VAE), which is competitive in both classification error and log-likelihood, despite optimizing both simultaneously and using a very simple encoder/decoder architecture.", "target": ["Combiniamo alberi decisionali differenziabili con autoencoder variazionali supervisionati per migliorare l'interpretabilità della classificazione.", "Questo articolo propone un modello ibrido di un autoencoder variazionale composto con un albero decisionale differenziabile, e uno schema di training di accompagnamento, con esperimenti che dimostrano le prestazioni di classificazione dell'albero, le prestazioni di negative log likelihood e l'interpretabilità dello spazio latente.", "L'articolo cerca di costruire un classificatore interpretabile e accurato impilando un VAE supervisionato e un albero decisionale differenziabile"]} +{"source": "We propose Regularized Learning under Label shifts (RLLS), a principled and a practical domain-adaptation algorithm to correct for shifts in the label distribution between a source and a target domain. We first estimate importance weights using labeled source data and unlabeled target data, and then train a classifier on the weighted source samples. We derive a generalization bound for the classifier on the target domain which is independent of the (ambient) data dimensions, and instead only depends on the complexity of the function class. To the best of our knowledge, this is the first generalization bound for the label-shift problem where the labels in the target domain are not available. Based on this bound, we propose a regularized estimator for the small-sample regime which accounts for the uncertainty in the estimated weights. Experiments on the CIFAR-10 and MNIST datasets show that RLLS improves classification accuracy, especially in the low sample and large-shift regimes, compared to previous methods.", "target": ["Un approccio pratico e garantito per il training di classificatori efficienti in presenza di spostamenti di label tra i dataset di origine e di destinazione", "Gli autori propongono un nuovo algoritmo per migliorare la stabilità della procedura di stima dell'importanza di una classe con una procedura a due fasi.", "Gli autori considerano il problema dell'apprendimento sotto label-shift, dove le proporzioni delle label differiscono mentre le condizioni sono uguali, e propongono uno stimatore migliorato con regolarizzazione."]} +{"source": "The statistics of the real visual world presents a long-tailed distribution: a few classes have significantly more training instances than the remaining classes in a dataset. This is because the real visual world has a few classes that are common while others are rare. Unfortunately, the performance of a convolutional neural network is typically unsatisfactory when trained using a long-tailed dataset. To alleviate this issue, we propose a method that discriminatively learns an embedding in which a simple Bayesian classifier can balance the class-priors to generalize well for rare classes. To this end, the proposed approach uses a Gaussian mixture model to factor out class-likelihoods and class-priors in a long-tailed dataset. The proposed method is simple and easy-to-implement in existing deep learning frameworks. Experiments on publicly available datasets show that the proposed approach improves the performance on classes with few training instances, while maintaining a comparable performance to the state-of-the-art on classes with abundant training examples.", "target": ["Approccio per migliorare la precisione di classificazione sulle classi nella coda della distribuzione.", "L'obiettivo principale di questo articolo è quello di imparare un classificatore basato su ConvNet che si comporta meglio per le classi nella coda della distribuzione delle occorrenze delle classi.", "Proposta di un framework bayesiano con un modello di mistura gaussiana per affrontare un problema nelle applicazioni di classificazione, cioè che il numero di dati di training da classi diverse è sbilanciato."]} +{"source": "As deep reinforcement learning is being applied to more and more tasks, there is a growing need to better understand and probe the learned agents. Visualizing and understanding the decision making process can be very valuable to comprehend and identify problems in the learned behavior. However, this topic has been relatively under-explored in the reinforcement learning community. In this work we present a method for synthesizing states of interest for a trained agent. Such states could be situations (e.g. crashing or damaging a car) in which specific actions are necessary. Further, critical states in which a very high or a very low reward can be achieved (e.g. risky states) are often interesting to understand the situational awareness of the system. To this end, we learn a generative model over the state space of the environment and use its latent space to optimize a target function for the state of interest. In our experiments we show that this method can generate insightful visualizations for a variety of environments and reinforcement learning methods. We explore these issues in the standard Atari benchmark games as well as in an autonomous driving simulator. Based on the efficiency with which we have been able to identify significant decision scenarios with this technique, we believe this general approach could serve as an important tool for AI safety applications.", "target": ["Presentiamo un metodo per sintetizzare gli stati di interesse per gli agenti di reinforcement learning al fine di analizzare il loro comportamento.", "Questo articolo propone un modello generativo di osservazioni visive in RL che è in grado di generare osservazioni di interesse.", "Un approccio per la visualizzazione degli stati di interesse che coinvolge un autoencoder variazionale che impara a ricostruire lo spazio degli stati e una fase di ottimizzazione che trova i parametri di condizionamento per generare immagini sintetiche."]} +{"source": "We introduce the deep abstaining classifier -- a deep neural network trained with a novel loss function that provides an abstention option during training. This allows the DNN to abstain on confusing or difficult-to-learn examples while improving performance on the non-abstained samples. We show that such deep abstaining classifiers can: (i) learn representations for structured noise -- where noisy training labels or confusing examples are correlated with underlying features -- and then learn to abstain based on such features; (ii) enable robust learning in the presence of arbitrary or unstructured noise by identifying noisy samples; and (iii) be used as an effective out-of-category detector that learns to reliably abstain when presented with samples from unknown classes. We provide analytical results on loss function behavior that enable automatic tuning of accuracy and coverage, and demonstrate the utility of the deep abstaining classifier using multiple image benchmarks, Results indicate significant improvement in learning in the presence of label noise.", "target": ["Una deep neural network di astensione addestrata con una nuova loss che impara rappresentazioni per quando astenersi permettendo un apprendimento robusto in presenza di diversi tipi di rumore.", "Una nuova loss per il training di una deep neural network che può astenersi, con prestazioni esaminate da angoli diversi in presenza di rumore strutturato e non strutturato, e rilevamento dell'open world.", "Questo manoscritto introduce classificatori deep con astensione che modificano la cross-entropy loss multiclasse con una loss di astensione, che viene poi applicata a task di classificazione di immagini perturbate"]} +{"source": "Temporal Difference learning with function approximation has been widely used recently and has led to several successful results. However, compared with the original tabular-based methods, one major drawback of temporal difference learning with neural networks and other function approximators is that they tend to over-generalize across temporally successive states, resulting in slow convergence and even instability. In this work, we propose a novel TD learning method, Hadamard product Regularized TD (HR-TD), that reduces over-generalization and thus leads to faster convergence. This approach can be easily applied to both linear and nonlinear function approximators. HR-TD is evaluated on several linear and nonlinear benchmark domains, where we show improvement in learning behavior and performance.", "target": ["Una tecnica di regolarizzazione per TD learning che evita la sovrageneralizzazione temporale, specialmente nelle deep network", "Una variazione su TD learning per il caso dell'approssimazione della funzione che tenta di risolvere il problema della sovra-generalizzazione attraverso stati temporalmente successivi.", "L'articolo introduce HR-TD, una variante dell'algoritmo TD(0), intesa a migliorare il problema della sovrageneralizzazione nel TD convenzionale"]} +{"source": "We present an efficient convolution kernel for Convolutional Neural Networks (CNNs) on unstructured grids using parameterized differential operators while focusing on spherical signals such as panorama images or planetary signals. To this end, we replace conventional convolution kernels with linear combinations of differential operators that are weighted by learnable parameters. Differential operators can be efficiently estimated on unstructured grids using one-ring neighbors, and learnable parameters can be optimized through standard back-propagation. As a result, we obtain extremely efficient neural networks that match or outperform state-of-the-art network architectures in terms of performance but with a significantly lower number of network parameters. We evaluate our algorithm in an extensive series of experiments on a variety of computer vision and climate science tasks, including shape classification, climate pattern segmentation, and omnidirectional image semantic segmentation. Overall, we present (1) a novel CNN approach on unstructured grids using parameterized differential operators for spherical signals, and (2) we show that our unique kernel parameterization allows our model to achieve the same or higher accuracy with significantly fewer network parameters.", "target": ["Presentiamo un nuovo kernel CNN per griglie non strutturate per segnali sferici, e mostriamo una precisione significativa e un guadagno di efficienza dei parametri su task come la classificazione 3D e la segmentazione omnidirezionale delle immagini.", "Un metodo efficiente che permette il deep learning su dati sferici che raggiunge numeri competitivi/stato dell'arte con molti meno parametri rispetto agli approcci popolari.", "L'articolo propone un nuovo kernel di convoluzione per CNN sulle grid non strutturate e formula la convoluzione con una combinazione lineare di operatori differenziali."]} +{"source": "Prediction is arguably one of the most basic functions of an intelligent system. In general, the problem of predicting events in the future or between two waypoints is exceedingly difficult. However, most phenomena naturally pass through relatively predictable bottlenecks---while we cannot predict the precise trajectory of a robot arm between being at rest and holding an object up, we can be certain that it must have picked the object up. To exploit this, we decouple visual prediction from a rigid notion of time. While conventional approaches predict frames at regularly spaced temporal intervals, our time-agnostic predictors (TAP) are not tied to specific times so that they may instead discover predictable \"bottleneck\" frames no matter when they occur. We evaluate our approach for future and intermediate frame prediction across three robotic manipulation tasks. Our predictions are not only of higher visual quality, but also correspond to coherent semantic subgoals in temporally extended tasks.", "target": ["Nei task di predizione visiva, lasciare che il modello predittivo scelga quali tempi predire fa due cose: (i) migliora la qualità della predizione e (ii) porta a predizioni semanticamente coerenti del \"bottleneck state\", che sono utili per la pianificazione.", "Un metodo per la predizione dei frame in un video, l'approccio comprende che la predizione dell'obiettivo è galleggiante, risolto da un minimo sull'errore di predizione.", "Riformula il task di predizione/interpolazione video in modo che un predittore non sia costretto a generare frame a intervalli di tempo fissi, ma sia invece addestrato a generare frame che accadono in qualsiasi punto del futuro."]} +{"source": "In cities with tall buildings, emergency responders need an accurate floor level location to find 911 callers quickly. We introduce a system to estimate a victim's floor level via their mobile device's sensor data in a two-step process. First, we train a neural network to determine when a smartphone enters or exits a building via GPS signal changes. Second, we use a barometer equipped smartphone to measure the change in barometric pressure from the entrance of the building to the victim's indoor location. Unlike impractical previous approaches, our system is the first that does not require the use of beacons, prior knowledge of the building infrastructure, or knowledge of user behavior. We demonstrate real-world feasibility through 63 experiments across five different tall buildings throughout New York City where our system predicted the correct floor level with 100% accuracy.", "target": ["Abbiamo usato una LSTM per rilevare quando uno smartphone entra in un edificio. Poi abbiamo previsto il piano in cui si trova il dispositivo utilizzando i dati dei sensori a bordo dello smartphone.", "L'articolo introduce un sistema per stimare il livello di un piano di un edificio attraverso i dati del sensore del loro dispositivo mobile utilizzando una LSTM e i cambiamenti nella pressione barometrica", "Proposta di un metodo in due fasi per determinare su quale piano si trova un telefono cellulare all'interno di un edificio alto."]} +{"source": "Sparse reward is one of the most challenging problems in reinforcement learning (RL). Hindsight Experience Replay (HER) attempts to address this issue by converting a failure experience to a successful one by relabeling the goals. Despite its effectiveness, HER has limited applicability because it lacks a compact and universal goal representation. We present Augmenting experienCe via TeacheR's adviCE (ACTRCE), an efficient reinforcement learning technique that extends the HER framework using natural language as the goal representation. We first analyze the differences among goal representation, and show that ACTRCE can efficiently solve difficult reinforcement learning problems in challenging 3D navigation tasks, whereas HER with non-language goal representation failed to learn. We also show that with language goal representations, the agent can generalize to unseen instructions, and even generalize to instructions with unseen lexicons. We further demonstrate it is crucial to use hindsight advice to solve challenging tasks, but we also found that little amount of hindsight advice is sufficient for the learning to take off, showing the practical aspect of the method.", "target": ["Combinare la rappresentazione degli obiettivi linguistici con i replay dell'esperienza a posteriori.", "Questo articolo considera l'assunzione implicita nell'experience replay a posteriori, cioè che vi sia accesso a una mappatura dagli stati agli obiettivi, e propone una rappresentazione degli obiettivi in linguaggio naturale.", "Questa presentazione utilizza il framework Hindsight Experience Replay con obiettivi in linguaggio naturale per migliorare la sample efficiency dei modelli instruction-following."]} +{"source": "Learning rich and compact representations is an open topic in many fields such as word embedding, visual question-answering, object recognition or image retrieval. Although deep neural networks (convolutional or not) have made a major breakthrough during the last few years by providing hierarchical, semantic and abstract representations for all of these tasks, these representations are not necessary as rich as needed nor as compact as expected. Models using higher order statistics, such as bilinear pooling, provide richer representations at the cost of higher dimensional features. Factorization schemes have been proposed but without being able to reach the original compactness of first order models, or at a heavy loss in performances. This paper addresses these two points by extending factorization schemes to codebook strategies, allowing compact representations with the same dimensionality as first order representations, but with second order performances. Moreover, we extend this framework with a joint codebook and factorization scheme, granting a reduction both in terms of parameters and computation cost. This formulation leads to state-of-the-art results and compact second-order models with few additional parameters and intermediate representations with a dimension similar to that of first-order statistics.", "target": ["Proponiamo un codebook congiunto e uno schema di fattorizzazione per migliorare il pooling di secondo ordine.", "Questo articolo presenta un modo per combinare le rappresentazioni fattorizzate di secondo ordine esistenti con un'assegnazione hard in stile codebook.", "Proposta di una nuova rappresentazione bilineare basata su un modello codebook, e una formulazione efficiente in cui le proiezioni basate su codebook sono fattorizzate tramite proiezione condivisa per ridurre ulteriormente il numero dei parametri."]} +{"source": "Natural language understanding research has recently shifted towards complex Machine Learning and Deep Learning algorithms. Such models often outperform their simpler counterparts significantly. However, their performance relies on the availability of large amounts of labeled data, which are rarely available. To tackle this problem, we propose a methodology for extending training datasets to arbitrarily big sizes and training complex, data-hungry models using weak supervision. We apply this methodology on biomedical relation extraction, a task where training datasets are excessively time-consuming and expensive to create, yet has a major impact on downstream applications such as drug discovery. We demonstrate in two small-scale controlled experiments that our method consistently enhances the performance of an LSTM network, with performance improvements comparable to hand-labeled training data. Finally, we discuss the optimal setting for applying weak supervision using this methodology.", "target": ["Proponiamo e applichiamo una metodologia di meta-learning basata sulla weak supervision, per combinare l'apprendimento semi-supervisionato ed ensemble sul task di estrazione di relazioni biomediche.", "Un metodo semi-supervised per la classificazione delle relazioni, che addestra più learner di base usando un piccolo dataset annotato e applica alcuni di loro per annotare esempi non annotati per l'apprendimento semi-supervised.", "Questo articolo affronta il problema della generazione di dati di training per l'estrazione di relazioni biologiche, e utilizza le predizioni dei dati annotati da classificatori deboli come dati di training aggiuntivi per un algoritmo di meta learning.", "Questo articolo propone una combinazione di semi-supervised e ensemble learning per l'estrazione di informazioni, con esperimenti condotti su un task di estrazione di relazioni biomediche"]} +{"source": "We introduce contextual explanation networks (CENs)---a class of models that learn to predict by generating and leveraging intermediate explanations. CENs are deep networks that generate parameters for context-specific probabilistic graphical models which are further used for prediction and play the role of explanations. Contrary to the existing post-hoc model-explanation tools, CENs learn to predict and to explain jointly. Our approach offers two major advantages: (i) for each prediction, valid instance-specific explanations are generated with no computational overhead and (ii) prediction via explanation acts as a regularization and boosts performance in low-resource settings. We prove that local approximations to the decision boundary of our networks are consistent with the generated explanations. Our results on image and text classification and survival analysis tasks demonstrate that CENs are competitive with the state-of-the-art while offering additional insights behind each prediction, valuable for decision support.", "target": ["Una classe di reti che generano modelli semplici al volo (chiamati explanation) che agiscono come un regolarizzatore e permettono una diagnostica coerente del modello e l'interpretabilità.", "Gli autori sostengono che lavori precedenti integrino direttamente le reti neurali nei graphical model come componenti, il che rende i modelli non interpretabili.", "Proposta di una combinazione di reti neurali e graphical model utilizzando una deep neural network per prevedere i parametri di un graphical model."]} +{"source": "The goal of imitation learning (IL) is to enable a learner to imitate an expert’s behavior given the expert’s demonstrations. Recently, generative adversarial imitation learning (GAIL) has successfully achieved it even on complex continuous control tasks. However, GAIL requires a huge number of interactions with environment during training. We believe that IL algorithm could be more applicable to the real-world environments if the number of interactions could be reduced. To this end, we propose a model free, off-policy IL algorithm for continuous control. The keys of our algorithm are two folds: 1) adopting deterministic policy that allows us to derive a novel type of policy gradient which we call deterministic policy imitation gradient (DPIG), 2) introducing a function which we call state screening function (SSF) to avoid noisy policy updates with states that are not typical of those appeared on the expert’s demonstrations. Experimental results show that our algorithm can achieve the goal of IL with at least tens of times less interactions than GAIL on a variety of continuous control tasks.", "target": ["Proponiamo un algoritmo di imitation learning senza modello che è in grado di ridurre il numero di interazioni con l'ambiente rispetto all'algoritmo di imitation learning allo stato dell'arte, cioè GAIL.", "Propone di estendere l'algoritmo deterministico del policy gradient per imparare dalle dimostrazioni, mentre viene combinato con un tipo di stima della densità dell'esperto.", "Questo articolo considera il problema dell'imitation learning model-free e propone un'estensione dell'algoritmo di generative adversarial imitation learning, sostituendo la policy stocastica del learner con una deterministica.", "L'articolo combina IRL, adversarial training e idee dai policy gradient deterministici con l'obiettivo di diminuire la sample complexity"]} +{"source": "Convolution acts as a local feature extractor in convolutional neural networks (CNNs). However, the convolution operation is not applicable when the input data is supported on an irregular graph such as with social networks, citation networks, or knowledge graphs. This paper proposes the topology adaptive graph convolutional network (TAGCN), a novel graph convolutional network that generalizes CNN architectures to graph-structured data and provides a systematic way to design a set of fixed-size learnable filters to perform convolutions on graphs. The topologies of these filters are adaptive to the topology of the graph when they scan the graph to perform convolution, replacing the square filter for the grid-structured data in traditional CNNs. The outputs are the weighted sum of these filters’ outputs, extraction of both vertex features and strength of correlation between vertices. It can be used with both directed and undirected graphs. The proposed TAGCN not only inherits the properties of convolutions in CNN for grid-structured data, but it is also consistent with convolution as defined in graph signal processing. Further, as no approximation to the convolution is needed, TAGCN exhibits better performance than existing graph-convolution-approximation methods on a number of data sets. As only the polynomials of degree two of the adjacency matrix are used, TAGCN is also computationally simpler than other recent methods.", "target": ["Graph CNN a bassa complessità computazionale (senza approssimazione) con una migliore accuratezza di classificazione", "Propone un nuovo approccio CNN alla classificazione dei grafi usando un filtro basato su cammini in uscita di lunghezza crescente per incorporare informazioni da vertici più distanti in un solo passo di propagazione.", "Proposta di una nuova architettura di rete neurale per la classificazione semi-supervisionata dei grafi, basandosi sui filtri polinomiali dei grafi e utilizzandoli su layer successivi di una rete neurale con funzioni di attivazione ReLU.", "L'articolo introduce Topology Adaptive GCN per generalizzare le reti di convoluzione ai dati strutturati a grafo"]} +{"source": "Inspired by the phenomenon of catastrophic forgetting, we investigate the learning dynamics of neural networks as they train on single classification tasks. Our goal is to understand whether a related phenomenon occurs when data does not undergo a clear distributional shift. We define a ``forgetting event'' to have occurred when an individual training example transitions from being classified correctly to incorrectly over the course of learning. Across several benchmark data sets, we find that: (i) certain examples are forgotten with high frequency, and some not at all; (ii) a data set's (un)forgettable examples generalize across neural architectures; and (iii) based on forgetting dynamics, a significant fraction of examples can be omitted from the training data set while still maintaining state-of-the-art generalization performance.", "target": ["Mostriamo che la catastrophic forgetting si verifica all'interno di quello che è considerato un singolo task e troviamo che gli esempi che non sono inclini alla dimenticanza possono essere rimossi dal training set senza perdita di generalizzazione.", "Il paper studia il comportamento di forgetting degli esempi di training durante SGD, e mostra che esistono \"esempi di supporto\" nel training delle reti neurali attraverso diverse architetture di rete.", "Questo articolo analizza la misura in cui le reti imparano a classificare correttamente esempi specifici e poi dimenticano questi esempi nel corso del training.", "L'articolo studia se alcuni esempi nel training delle reti neurali sono più difficili da imparare di altri. Tali esempi sono dimenticati e ri-appresi più volte attraverso l'apprendimento."]} +{"source": "Discovering objects and their attributes is of great importance for autonomous agents to effectively operate in human environments. This task is particularly challenging due to the ubiquitousness of objects and all their nuances in perceptual and semantic detail. In this paper we present an unsupervised approach for learning disentangled representations of objects entirely from unlabeled monocular videos. These continuous representations are not biased by or limited by a discrete set of labels determined by human labelers. The proposed representation is trained with a metric learning loss, where objects with homogeneous features are pushed together, while those with heterogeneous features are pulled apart. We show these unsupervised embeddings allow to discover object attributes and can enable robots to self-supervise in previously unseen environments. We quantitatively evaluate performance on a large-scale synthetic dataset with 12k object models, as well as on a real dataset collected by a robot and show that our unsupervised object understanding generalizes to previously unseen objects. Specifically, we demonstrate the effectiveness of our approach on robotic manipulation tasks, such as pointing at and grasping of objects. An interesting and perhaps surprising finding in this approach is that given a limited set of objects, object correspondences will naturally emerge when using metric learning without requiring explicit positive pairs.", "target": ["Un approccio unsupervised per l'apprendimento di rappresentazioni disentangled di oggetti interamente da video monoculari senza label.", "Progetta una rappresentazione delle feature da sequenze video catturate da una scena da diversi punti di vista.", "Proposta di un metodo di representation learning unsupervised per gli input visivi che incorpora un approccio di metric learning che avvicina le coppie di patch di immagini che sono nearest neighbor nello spazio di embedding mentre spinge a parte altre coppie.", "Questo articolo esplora il self-supervised learning di rappresentazioni di oggetti, con l'idea principale di incoraggiare gli oggetti con feature simili ad essere ulteriormente \"attratti\" l'uno dall'altro."]} +{"source": "Learning from a scalar reward in continuous action space environments is difficult and often requires millions if not billions of interactions. We introduce state aligned vector rewards, which are easily defined in metric state spaces and allow our deep reinforcement learning agent to tackle the curse of dimensionality. Our agent learns to map from action distributions to state change distributions implicitly defined in a quantile function neural network. We further introduce a new reinforcement learning technique inspired by quantile regression which does not limit agents to explicitly parameterized action distributions. Our results in high dimensional state spaces show that training with vector rewards allows our agent to learn multiple times faster than an agent training with scalar rewards.", "target": ["Si esegue il training di un agente che predice i cambiamenti di stato dalle distribuzioni delle azioni tramite un vettore di reward allineato allo stato, usando una nuova tecnica di reinforcement learning ispirata alla regressione quantile.", "Presenta un algoritmo che mira ad accelerare il reinforcement learning in situazioni in cui la reward è allineata con lo spazio di stato.", "Questo articolo affronta il RL nello spazio d'azione continuo, utilizzando una policy ri-parametrizzata e un nuovo obiettivo di training basato su vettori.", "Questo lavoro propone di mescolare il RL distributivo con una rete incaricata di modellare l'evoluzione del mondo in termini di quantili, sostenendo miglioramenti nella sample efficiency."]} +{"source": "We propose Episodic Backward Update - a new algorithm to boost the performance of a deep reinforcement learning agent by fast reward propagation. In contrast to the conventional use of the replay memory with uniform random sampling, our agent samples a whole episode and successively propagates the value of a state into its previous states. Our computationally efficient recursive algorithm allows sparse and delayed rewards to propagate effectively throughout the sampled episode. We evaluate our algorithm on 2D MNIST Maze Environment and 49 games of the Atari 2600 Environment and show that our agent improves sample efficiency with a competitive computational cost.", "target": ["Proponiamo Episodic Backward Update, un nuovo algoritmo di deep reinforcement learning che campiona le transizioni episodio per episodio e aggiorna i valori in modo ricorsivo all'indietro per ottenere un apprendimento veloce e stabile.", "Propone una nuova DQN in cui gli obiettivi sono calcolati su un episodio completo con un aggiornamento all'indietro (dalla fine all'inizio) per una propagazione più veloce delle reward dalla fine dell'episodio.", "Gli autori propongono di modificare l'algoritmo DQN applicando l'operatore max di Bellman ricorsivamente su una traiettoria con un certo decadimento per evitare di accumulare errori con il max innestato.", "Nelle deep-Q network, si aggiornano i valori Q a partire dalla fine dell'episodio per facilitare la propagazione rapida delle reward lungo l'episodio."]} +{"source": "Survival Analysis (time-to-event analysis) in the presence of multiple possible adverse events, i.e., competing risks, is a challenging, yet very important problem in medicine, finance, manufacturing, etc. Extending classical survival analysis to competing risks is not trivial since only one event (e.g. one cause of death) is observed and hence, the incidence of an event of interest is often obscured by other related competing events. This leads to the nonidentifiability of the event times’ distribution parameters, which makes the problem significantly more challenging. In this work we introduce Siamese Survival Prognosis Network, a novel Siamese Deep Neural Network architecture that is able to effectively learn from data in the presence of multiple adverse events. The Siamese Survival Network is especially crafted to issue pairwise concordant time-dependent risks, in which longer event times are assigned lower risks. Furthermore, our architecture is able to directly optimize an approximation to the C-discrimination index, rather than relying on well-known metrics of cross-entropy etc., and which are not able to capture the unique requirements of survival analysis with competing risks. Our results show consistent performance improvements on a number of publicly available medical datasets over both statistical and deep learning state-of-the-art methods.", "target": ["In questo lavoro introduciamo una nuova architettura di deep neural network siamese che è in grado di imparare efficacemente dai dati in presenza di più eventi avversi.", "Questo articolo introduce le reti neurali siamesi nel framework dei rischi concorrenti ottimizzando direttamente l'indice c", "Gli autori affrontano i problemi di stima del rischio in un contesto di analisi della sopravvivenza con rischi concorrenti e propongono di ottimizzare direttamente l'indice di discriminazione dipendente dal tempo utilizzando una siamese survival network"]} +{"source": "The digitization of data has resulted in making datasets available to millions of users in the form of relational databases and spreadsheet tables. However, a majority of these users come from diverse backgrounds and lack the programming expertise to query and analyze such tables. We present a system that allows for querying data tables using natural language questions, where the system translates the question into an executable SQL query. We use a deep sequence to sequence model in wich the decoder uses a simple type system of SQL expressions to structure the output prediction. Based on the type, the decoder either copies an output token from the input question using an attention-based copying mechanism or generates it from a fixed vocabulary. We also introduce a value-based loss function that transforms a distribution over locations to copy from into a distribution over the set of input tokens to improve training of our model. We evaluate our model on the recently released WikiSQL dataset and show that our model trained using only supervised learning significantly outperforms the current state-of-the-art Seq2SQL model that uses reinforcement learning.", "target": ["Presentiamo un modello di pointer network type-based insieme a un metodo di loss basato sul valore per addestrare efficacemente un modello neurale per tradurre il linguaggio naturale in SQL.", "L'articolo sostiene di sviluppare un nuovo metodo per mappare le query del linguaggio naturale in SQL utilizzando una grammatica per guidare la decodifica e utilizzando una nuova loss per il meccanismo di pointer/copy"]} +{"source": "To backpropagate the gradients through stochastic binary layers, we propose the augment-REINFORCE-merge (ARM) estimator that is unbiased, exhibits low variance, and has low computational complexity. Exploiting variable augmentation, REINFORCE, and reparameterization, the ARM estimator achieves adaptive variance reduction for Monte Carlo integration by merging two expectations via common random numbers. The variance-reduction mechanism of the ARM estimator can also be attributed to either antithetic sampling in an augmented space, or the use of an optimal anti-symmetric \"self-control\" baseline function together with the REINFORCE estimator in that augmented space. Experimental results show the ARM estimator provides state-of-the-art performance in auto-encoding variational inference and maximum likelihood estimation, for discrete latent variable models with one or multiple stochastic binary layers. Python code for reproducible research is publicly available.", "target": ["Uno stimatore di gradiente unbiased e a bassa varianza per modelli di variabili latenti discreti", "Propone una nuova tecnica di riduzione della varianza da usare quando si calcola un gradiente di expected loss dove l'aspettativa è rispetto a variabili casuali binarie indipendenti.", "Un algoritmo che combina Rao-Blackwellization e numeri casuali comuni per abbassare la varianza dello stimatore del gradiente della funzione di punteggio nel caso speciale delle reti binarie stocastiche", "Uno stimatore unbiased e a bassa varianza augment-REINFORCE-merge (ARM) per il calcolo e la backpropagation dei gradienti nelle reti neurali binarie"]} +{"source": "Mini-batch stochastic gradient descent (SGD) is state of the art in large scale distributed training. The scheme can reach a linear speed-up with respect to the number of workers, but this is rarely seen in practice as the scheme often suffers from large network delays and bandwidth limits. To overcome this communication bottleneck recent works propose to reduce the communication frequency. An algorithm of this type is local SGD that runs SGD independently in parallel on different workers and averages the sequences only once in a while. This scheme shows promising results in practice, but eluded thorough theoretical analysis. We prove concise convergence rates for local SGD on convex problems and show that it converges at the same rate as mini-batch SGD in terms of number of evaluated gradients, that is, the scheme achieves linear speed-up in the number of workers and mini-batch size. The number of communication rounds can be reduced up to a factor of T^{1/2}---where T denotes the number of total steps---compared to mini-batch SGD. This also holds for asynchronous implementations. Local SGD can also be used for large scale training of deep learning models. The results shown here aim serving as a guideline to further explore the theoretical and practical aspects of local SGD in these applications.", "target": ["Dimostriamo che SGD locale parallelo raggiunge uno speedup lineare con una comunicazione molto minore rispetto a SGD parallelo in mini-batch.", "Fornisce una prova di convergenza per SGD locale, e dimostra che SGD locale può fornire gli stessi guadagni di accelerazione del minibatch, ma può essere in grado di comunicare molto meno.", "Questo articolo presenta un'analisi di SGD locale e dei bound sulla frequenza con cui gli stimatori ottenuti dall'esecuzione di SGD devono essere mediati per ottenere un aumento lineare della velocità di parallelizzazione.", "Gli autori analizzano l'algoritmo SGD locale, dove vengono eseguite $K$ catene parallele di SGD, e le iterazioni sono occasionalmente sincronizzate tra le macchine facendo la media"]} +{"source": "Extracting relevant information, causally inferring and predicting the future states with high accuracy is a crucial task for modeling complex systems. The endeavor to address these tasks is made even more challenging when we have to deal with high-dimensional heterogeneous data streams. Such data streams often have higher-order inter-dependencies across spatial and temporal dimensions. We propose to perform a soft-clustering of the data and learn its dynamics to produce a compact dynamical model while still ensuring the original objectives of causal inference and accurate predictions. To efficiently and rigorously process the dynamics of soft-clustering, we advocate for an information theory inspired approach that incorporates stochastic calculus and seeks to determine a trade-off between the predictive accuracy and compactness of the mathematical representation. We cast the model construction as a maximization of the compression of the state variables such that the predictive ability and causal interdependence (relatedness) constraints between the original data streams and the compact model are closely bounded. We provide theoretical guarantees concerning the convergence of the proposed learning algorithm. To further test the proposed framework, we consider a high-dimensional Gaussian case study and describe an iterative scheme for updating the new model parameters. Using numerical experiments, we demonstrate the benefits on compression and prediction accuracy for a class of dynamical systems. Finally, we apply the proposed algorithm to the real-world dataset of multimodal sentiment intensity and show improvements in prediction with reduced dimensions.", "target": ["Percezione compatta del processo dinamico", "Studia il problema di rappresentare in modo compatto il modello di un sistema dinamico complesso preservando l'informazione usando un metodo di information bottleneck.", "Questo articolo studia la dinamica lineare gaussiana e propone un algoritmo per calcolare l'Information Bottleneck Hierarchy (IBH)."]} +{"source": "We propose the dense RNN, which has the fully connections from each hidden state to multiple preceding hidden states of all layers directly. As the density of the connection increases, the number of paths through which the gradient flows can be increased. It increases the magnitude of gradients, which help to prevent the vanishing gradient problem in time. Larger gradients, however, can also cause exploding gradient problem. To complement the trade-off between two problems, we propose an attention gate, which controls the amounts of gradient flows. We describe the relation between the attention gate and the gradient flows by approximation. The experiment on the language modeling using Penn Treebank corpus shows dense connections with the attention gate improve the model’s performance.", "target": ["RNN densa che ha connessioni complete da ogni stato nascosto direttamente a più stati nascosti precedenti di tutti i layer.", "Propone una nuova architettura RNN che modella meglio le dipendenze a lungo termine, può imparare la rappresentazione multiscala dei dati sequenziali e aggirare il problema dei gradienti usando unità di gating parametrizzate.", "Questo articolo propone un'architettura RNN densa fully connected con connessioni gated per ogni layer e connessioni del layer precedente, e i suoi risultati sul task di modellazione a livello di caratteri PTB."]} +{"source": "We propose a new algorithm for training generative adversarial networks to jointly learn latent codes for both identities (e.g. individual humans) and observations (e.g. specific photographs). In practice, this means that by fixing the identity portion of latent codes, we can generate diverse images of the same subject, and by fixing the observation portion we can traverse the manifold of subjects while maintaining contingent aspects such as lighting and pose. Our algorithm features a pairwise training scheme in which each sample from the generator consists of two images with a common identity code. Corresponding samples from the real dataset consist of two distinct photographs of the same subject. In order to fool the discriminator, the generator must produce images that are both photorealistic, distinct, and appear to depict the same person. We augment both the DCGAN and BEGAN approaches with Siamese discriminators to accommodate pairwise training. Experiments with human judges and an off-the-shelf face verification system demonstrate our algorithm’s ability to generate convincing, identity-matched photographs.", "target": ["Le SD-GAN distinguono i codici latenti in base alle comunanze note in un dataset (ad esempio, fotografie che ritraggono la stessa persona).", "Questo articolo indaga il problema della generazione controllata di immagini e propone un algoritmo che produce una coppia di immagini con la stessa identità.", "Questo articolo propone SD-GAN, un metodo di training delle GAN per distinguere le informazioni di identità e non identità nel vettore latente di input Z."]} +{"source": "The goal of unpaired cross-domain translation is to learn useful mappings between two domains, given unpaired sets of datapoints from these domains. While this formulation is highly underconstrained, recent work has shown that it is possible to learn mappings useful for downstream tasks by encouraging approximate cycle consistency in the mappings between the two domains [Zhu et al., 2017]. In this work, we propose AlignFlow, a framework for unpaired cross-domain translation that ensures exact cycle consistency in the learned mappings. Our framework uses a normalizing flow model to specify a single invertible mapping between the two domains. In contrast to prior works in cycle-consistent translations, we can learn AlignFlow via adversarial training, maximum likelihood estimation, or a hybrid of the two methods. Theoretically, we derive consistency results for AlignFlow which guarantee recovery of desirable mappings under suitable assumptions. Empirically, AlignFlow demonstrates significant improvements over relevant baselines on image-to-image translation and unsupervised domain adaptation tasks on benchmark datasets.", "target": ["Proponiamo un learning framework per le traduzioni cross-domain che è esattamente cycle-consistent e può essere appreso tramite adversarial training, stima di massima verosimiglianza, o un ibrido dei due.", "Propone AlignFlow, un modo efficiente di implementare il principio di coerenza del ciclo usando flussi invertibili.", "Modelli di flusso per la traduzione da immagine a immagine non accoppiata"]} +{"source": "Program synthesis is a class of regression problems where one seeks a solution, in the form of a source-code program, that maps the inputs to their corresponding outputs exactly. Due to its precise and combinatorial nature, it is commonly formulated as a constraint satisfaction problem, where input-output examples are expressed constraints, and solved with a constraint solver. A key challenge of this formulation is that of scalability: While constraint solvers work well with few well-chosen examples, constraining the entire set of example constitutes a significant overhead in both time and memory. In this paper we address this challenge by constructing a representative subset of examples that is both small and is able to constrain the solver sufficiently. We build the subset one example at a time, using a trained discriminator to predict the probability of unchosen input-output examples conditioned on the chosen input-output examples, adding the least probable example to the subset. Experiment on a diagram drawing domain shows our approach produces subset of examples that are small and representative for the constraint solver.", "target": ["In un contesto di program synthesis in cui l'input è un insieme di esempi, riduciamo il costo calcolando un sottoinsieme di esempi rappresentativi", "Propone un metodo per identificare esempi rappresentativi per program synthesis per aumentare la scalabilità delle soluzioni di programmazione vincolata esistenti.", "Un metodo per scegliere un sottoinsieme di esempi su cui eseguire un risolutore di vincoli per risolvere problemi di program synthesis.", "Questo articolo propone un metodo per accelerare i sintetizzatori di programmi general-purpose."]} +{"source": "Humans possess an ability to abstractly reason about objects and their interactions, an ability not shared with state-of-the-art deep learning models. Relational networks, introduced by Santoro et al. (2017), add the capacity for relational reasoning to deep neural networks, but are limited in the complexity of the reasoning tasks they can address. We introduce recurrent relational networks which increase the suite of solvable tasks to those that require an order of magnitude more steps of relational reasoning. We use recurrent relational networks to solve Sudoku puzzles and achieve state-of-the-art results by solving 96.6% of the hardest Sudoku puzzles, where relational networks fail to solve any. We also apply our model to the BaBi textual QA dataset solving 19/20 tasks which is competitive with state-of-the-art sparse differentiable neural computers. The recurrent relational network is a general purpose module that can augment any neural network model with the capacity to do many-step relational reasoning.", "target": ["Introduciamo Recurrent Relational Networks, un modulo di rete neurale potente e generale per il ragionamento relazionale, e lo usiamo per risolvere il 96,6% dei task Sudoku più difficili e 19/20 dei task BaBi.", "Introdotta la rete relazionale ricorrente (RRNs) che può essere aggiunta a qualsiasi rete neurale per aggiungere capacità di ragionamento relazionale.", "Introduzione di una deep neural network per structured prediction che raggiunge prestazioni allo stato dell'arte sui puzzle Soduku e sul task BaBi.", "Questo articolo descrive un metodo chiamato rete relazionale per aggiungere capacità di ragionamento relazionale alle deep neural network."]} +{"source": "Empirical risk minimization (ERM), with proper loss function and regularization, is the common practice of supervised classification. In this paper, we study training arbitrary (from linear to deep) binary classifier from only unlabeled (U) data by ERM. We prove that it is impossible to estimate the risk of an arbitrary binary classifier in an unbiased manner given a single set of U data, but it becomes possible given two sets of U data with different class priors. These two facts answer a fundamental question---what the minimal supervision is for training any binary classifier from only U data. Following these findings, we propose an ERM-based learning method from two sets of U data, and then prove it is consistent. Experiments demonstrate the proposed method could train deep models and outperform state-of-the-art methods for learning from two sets of U data.", "target": ["Tre prior di classe sono tutti necessari per addestrare deep model da soli dati U, mentre due qualsiasi non dovrebbero essere sufficienti.", "Propone uno stimatore unbiased che permette di addestrare modelli con supervisione debole su due dataset non annotati con prior di classe noti e discute le proprietà teoriche degli stimatori.", "Una metodologia per il training di qualsiasi classificatore binario da soli dati non annotati, e un metodo di empirical risk minimization per due insiemi di dati non annotati in cui sono dati i prior di classe."]} +{"source": "Deep Neural Networks (DNNs) excel on many complex perceptual tasks but it has proven notoriously difficult to understand how they reach their decisions. We here introduce a high-performance DNN architecture on ImageNet whose decisions are considerably easier to explain. Our model, a simple variant of the ResNet-50 architecture called BagNet, classifies an image based on the occurrences of small local image features without taking into account their spatial ordering. This strategy is closely related to the bag-of-feature (BoF) models popular before the onset of deep learning and reaches a surprisingly high accuracy on ImageNet (87.6% top-5 for 32 x 32 px features and Alexnet performance for 16 x16 px features). The constraint on local features makes it straight-forward to analyse how exactly each part of the image influences the classification. Furthermore, the BagNets behave similar to state-of-the art deep neural networks such as VGG-16, ResNet-152 or DenseNet-169 in terms of feature sensitivity, error distribution and interactions between image parts. This suggests that the improvements of DNNs over previous bag-of-feature classifiers in the last few years is mostly achieved by better fine-tuning rather than by qualitatively different decision strategies.", "target": ["L'aggregazione di prove di classe da molte piccole patch di immagini è sufficiente per risolvere ImageNet, produce modelli più interpretabili e può spiegare aspetti del processo decisionale delle DNN popolari.", "Questo articolo suggerisce un'architettura di rete neurale nuova e compatta che utilizza le informazioni all'interno delle feature di bag-of-words. L'algoritmo proposto utilizza solo le informazioni delle patch in modo indipendente ed esegue il voto di maggioranza utilizzando patch classificate in modo indipendente."]} +{"source": "Somatic cancer mutation detection at ultra-low variant allele frequencies (VAFs) is an unmet challenge that is intractable with current state-of-the-art mutation calling methods. Specifically, the limit of VAF detection is closely related to the depth of coverage, due to the requirement of multiple supporting reads in extant methods, precluding the detection of mutations at VAFs that are orders of magnitude lower than the depth of coverage. Nevertheless, the ability to detect cancer-associated mutations in ultra low VAFs is a fundamental requirement for low-tumor burden cancer diagnostics applications such as early detection, monitoring, and therapy nomination using liquid biopsy methods (cell-free DNA). Here we defined a spatial representation of sequencing information adapted for convolutional architecture that enables variant detection at VAFs, in a manner independent of the depth of sequencing. This method enables the detection of cancer mutations even in VAFs as low as 10x-4^, >2 orders of magnitude below the current state-of-the-art. We validated our method on both simulated plasma and on clinical cfDNA plasma samples from cancer patients and non-cancer controls. This method introduces a new domain within bioinformatics and personalized medicine – somatic whole genome mutation calling for liquid biopsy.", "target": ["Gli attuali metodi di mutazione somatica non funzionano con le biopsie liquide (cioè il sequenziamento a bassa copertura), applichiamo un'architettura CNN a una rappresentazione unica di una lettura e il suo alignment, mostriamo un miglioramento significativo rispetto ai metodi precedenti nel setting a bassa frequenza.", "Propone una soluzione basata su CNN chiamata Kittyhawk per somatic mutation calling a frequenze alleliche ultra basse.", "Un nuovo algoritmo per rilevare le mutazioni del cancro dal sequenziamento del DNA libero delle cellule che identificherà il contesto di sequenza che distingue gli errori di sequenziamento dalle vere mutazioni.", "Questo articolo propone un framework di deep learning per prevedere le mutazioni somatiche a frequenze estremamente basse che si verificano nel rilevamento del tumore dal DNA privo di cellule"]} +{"source": "This paper presents the formal release of {\\em MedMentions}, a new manually annotated resource for the recognition of biomedical concepts. What distinguishes MedMentions from other annotated biomedical corpora is its size (over 4,000 abstracts and over 350,000 linked mentions), as well as the size of the concept ontology (over 3 million concepts from UMLS 2017) and its broad coverage of biomedical disciplines. In addition to the full corpus, a sub-corpus of MedMentions is also presented, comprising annotations for a subset of UMLS 2017 targeted towards document retrieval. To encourage research in Biomedical Named Entity Recognition and Linking, data splits for training and testing are included in the release, and a baseline model and its metrics for entity linking are also described.", "target": ["L'articolo introduce un nuovo corpus gold-standard di letteratura scientifica biomedica annotato manualmente con le menzioni dei concetti UMLS.", "Dettaglia la costruzione di un dataset annotato manualmente che copre concetti biomedici che è più grande e coperto da un'ontologia più grande dei dataset precedenti.", "Questo articolo utilizza MedMentions, un modello TaggerOne semi-Markov per il concept recognition e linking end-to-end su un insieme di abstract di Pubmed per etichettare i documenti con concetti/entità biomediche"]} +{"source": "In this paper we propose a Deep Autoencoder Mixture Clustering (DAMIC) algorithm. It is based on a mixture of deep autoencoders where each cluster is represented by an autoencoder. A clustering network transforms the data into another space and then selects one of the clusters. Next, the autoencoder associated with this cluster is used to reconstruct the data-point. The clustering algorithm jointly learns the nonlinear data representation and the set of autoencoders. The optimal clustering is found by minimizing the reconstruction loss of the mixture of autoencoder network. Unlike other deep clustering algorithms, no regularization term is needed to avoid data collapsing to a single point. Our experimental evaluations on image and text corpora show significant improvement over state-of-the-art methods.", "target": ["Proponiamo un metodo di deep clustering dove al posto di un centroide ogni cluster è rappresentato da un autoencoder", "Presenta un deep clustering basato su una miscela di autoencoder, dove i datapoint sono assegnati a un cluster basato sull'errore di rappresentazione se la rete di autoencoder fosse usata per rappresentarlo.", "Un approccio di deep clustering che utilizza un framework con autoencoder per imparare un'embedding bidimensionale dei dati mentre si raggruppano i dati utilizzando una deep neural network.", "Un metodo di deep clustering che rappresenta ogni cluster con diversi autoencoder, funziona in modo end-to-end, e può anche essere usato per clusterizzare nuovi dati in arrivo senza rifare l'intera procedura di clustering."]} +{"source": "We propose a new Integral Probability Metric (IPM) between distributions: the Sobolev IPM. The Sobolev IPM compares the mean discrepancy of two distributions for functions (critic) restricted to a Sobolev ball defined with respect to a dominant measure mu. We show that the Sobolev IPM compares two distributions in high dimensions based on weighted conditional Cumulative Distribution Functions (CDF) of each coordinate on a leave one out basis. The Dominant measure mu plays a crucial role as it defines the support on which conditional CDFs are compared. Sobolev IPM can be seen as an extension of the one dimensional Von-Mises Cramer statistics to high dimensional distributions. We show how Sobolev IPM can be used to train Generative Adversarial Networks (GANs). We then exploit the intrinsic conditioning implied by Sobolev IPM in text generation. Finally we show that a variant of Sobolev GAN achieves competitive results in semi-supervised learning on CIFAR-10, thanks to the smoothness enforced on the critic by Sobolev GAN which relates to Laplacian regularization.", "target": ["Definiamo una nuova metrica di probabilità integrale (Sobolev IPM) e mostriamo come può essere utilizzata per il training di GAN per la generazione di testi e l'apprendimento semi-supervisionato.", "Suggerisce un nuovo schema di regolarizzazione per le GAN basato sulla norma di Sobolev, che misura le deviazioni tra le norme L2 delle derivate.", "Gli autori forniscono un altro tipo di GAN usando il setting tipico di una GAN ma con una classe di funzione diversa, e forniscono una procedura per addestrare le GAN con quel tipo di classe di funzione.", "L'articolo propone una diversa penalità del gradiente per i GAN critic che forza la norma quadratica attesa del gradiente ad essere uguale a 1"]} +{"source": "We propose in this paper a new approach to train the Generative Adversarial Nets (GANs) with a mixture of generators to overcome the mode collapsing problem. The main intuition is to employ multiple generators, instead of using a single one as in the original GAN. The idea is simple, yet proven to be extremely effective at covering diverse data modes, easily overcoming the mode collapsing problem and delivering state-of-the-art results. A minimax formulation was able to establish among a classifier, a discriminator, and a set of generators in a similar spirit with GAN. Generators create samples that are intended to come from the same distribution as the training data, whilst the discriminator determines whether samples are true data or generated by generators, and the classifier specifies which generator a sample comes from. The distinguishing feature is that internal samples are created from multiple generators, and then one of them will be randomly selected as final output similar to the mechanism of a probabilistic mixture model. We term our method Mixture Generative Adversarial Nets (MGAN). We develop theoretical analysis to prove that, at the equilibrium, the Jensen-Shannon divergence (JSD) between the mixture of generators’ distributions and the empirical data distribution is minimal, whilst the JSD among generators’ distributions is maximal, hence effectively avoiding the mode collapsing problem. By utilizing parameter sharing, our proposed model adds minimal computational cost to the standard GAN, and thus can also efficiently scale to large-scale datasets. We conduct extensive experiments on synthetic 2D data and natural image databases (CIFAR-10, STL-10 and ImageNet) to demonstrate the superior performance of our MGAN in achieving state-of-the-art Inception scores over latest baselines, generating diverse and appealing recognizable objects at different resolutions, and specializing in capturing different types of objects by the generators.", "target": ["Proponiamo un nuovo approccio per addestrare le GAN con una mistura di generatori per superare il problema del mode collapse.", "Affrontare il problema del mode collapse nelle GAN utilizzando una mistura vincolata di distribuzioni per il generatore e un classificatore ausiliario che predice la componente d'origine della mistura.", "L'articolo propone una mistura di generatori per addestrare le GAN senza costi computazionali aggiuntivi", "Gli autori presentano l'uso di MGAN, che mira a superare il problema del mode collapse mediante misture di generatori e che raggiunge risultati allo stato dell'arte"]} +{"source": "Allowing humans to interactively train artificial agents to understand language instructions is desirable for both practical and scientific reasons. Though, given the lack of sample efficiency in current learning methods, reaching this goal may require substantial research efforts. We introduce the BabyAI research platform, with the goal of supporting investigations towards including humans in the loop for grounded language learning. The BabyAI platform comprises an extensible suite of 19 levels of increasing difficulty. Each level gradually leads the agent towards acquiring a combinatorially rich synthetic language, which is a proper subset of English. The platform also provides a hand-crafted bot agent, which simulates a human teacher. We report estimated amount of supervision required for training neural reinforcement and behavioral-cloning agents on some BabyAI levels. We put forward strong evidence that current deep learning methods are not yet sufficiently sample-efficient in the context of learning a language with compositional properties.", "target": ["Presentiamo la piattaforma BabyAI per studiare l'efficienza dei dati dell'apprendimento del linguaggio con human in the loop", "Presenta una piattaforma di ricerca con un bot nel loop per imparare a eseguire istruzioni linguistiche in cui il linguaggio ha strutture compositive", "Introduce una piattaforma per l'apprendimento delle lingue grounded che sostituisce qualsiasi human in the loop con un insegnante euristico e utilizza una lingua sintetica mappata in un 2D grid world"]} +{"source": "Recently, there has been growing interest in methods that perform neural network compression, namely techniques that attempt to substantially reduce the size of a neural network without significant reduction in performance. However, most existing methods are post-processing approaches in that they take a learned neural network as input and output a compressed network by either forcing several parameters to take the same value (parameter tying via quantization) or pruning irrelevant edges (pruning) or both. In this paper, we propose a novel algorithm that jointly learns and compresses a neural network. The key idea in our approach is to change the optimization criteria by adding $k$ independent Gaussian priors over the parameters and a sparsity penalty. We show that our approach is easy to implement using existing neural network libraries, generalizes L1 and L2 regularization and elegantly enforces parameter tying as well as pruning constraints. Experimentally, we demonstrate that our new algorithm yields state-of-the-art compression on several standard benchmarks with minimal loss in accuracy while requiring little to no hyperparameter tuning as compared with related, competing approaches.", "target": ["Un prior k-means combinato con la regolarizzazione L1 produce risultati di compressione allo stato dell'arte.", "Questo articolo esplora la soft parameter tying e la compressione di DNN/CNN"]} +{"source": "The application of stochastic variance reduction to optimization has shown remarkable recent theoretical and practical success. The applicability of these techniques to the hard non-convex optimization problems encountered during training of modern deep neural networks is an open problem. We show that naive application of the SVRG technique and related approaches fail, and explore why.", "target": ["Il metodo SVRG fallisce sui moderni problemi di deep learning", "Questo articolo presenta un'analisi dei metodi in stile SVRG, mostrando che dropout, batch norm, data augmention (crop/rotazioni/traslazioni casuali) tendono ad aumentare il bias e/o la varianza degli update.", "Questo articolo studia l'applicabilità di SVGD alle moderne reti neurali e mostra che la semplice applicazione di SVGD tipicamente fallisce."]} +{"source": "The ground-breaking performance obtained by deep convolutional neural networks (CNNs) for image processing tasks is inspiring research efforts attempting to extend it for 3D geometric tasks. One of the main challenge in applying CNNs to 3D shape analysis is how to define a natural convolution operator on non-euclidean surfaces. In this paper, we present a method for applying deep learning to 3D surfaces using their spherical descriptors and alt-az anisotropic convolution on 2-sphere. A cascade set of geodesic disk filters rotate on the 2-sphere and collect spherical patterns and so to extract geometric features for various 3D shape analysis tasks. We demonstrate theoretically and experimentally that our proposed method has the possibility to bridge the gap between 2D images and 3D shapes with the desired rotation equivariance/invariance, and its effectiveness is evaluated in applications of non-rigid/ rigid shape classification and shape retrieval.", "target": ["Un metodo per applicare il deep learning alle superfici 3D usando i loro descrittori sferici e la convoluzione anisotropa alt-az su 2-sfere.", "Presenta uno schema di convoluzione polare anisotropa su una sfera unitaria sostituendo la traslazione del filtro con la rotazione del filtro.", "Questo articolo esplora il deep learning di forme 3D utilizzando la convoluzione anisotropa a 2 sfere di alt-az"]} +{"source": "Recent breakthroughs in computer vision make use of large deep neural networks, utilizing the substantial speedup offered by GPUs. For applications running on limited hardware, however, high precision real-time processing can still be a challenge. One approach to solving this problem is training networks with binary or ternary weights, thus removing the need to calculate multiplications and significantly reducing memory size. In this work, we introduce LR-nets (Local reparameterization networks), a new method for training neural networks with discrete weights using stochastic parameters. We show how a simple modification to the local reparameterization trick, previously used to train Gaussian distributed weights, enables the training of discrete weights. Using the proposed training we test both binary and ternary models on MNIST, CIFAR-10 and ImageNet benchmarks and reach state-of-the-art results on most experiments.", "target": ["Il training di reti binarie/ternarie usando la riparametrizzazione locale con l'approssimazione CLT", "Addestra le reti di distribuzione dei pesi binarie e ternarie usando la backpropagation per campionare le preattivazioni dei neuroni con il reparameterization trick", "Questo articolo suggerisce l'uso di parametri stocastici in combinazione con il reparameterization trick locale per addestrare reti neurali con pesi binari o ternari, che porta a risultati allo stato dell'arte."]} +{"source": "We present Optimal Completion Distillation (OCD), a training procedure for optimizing sequence to sequence models based on edit distance. OCD is efficient, has no hyper-parameters of its own, and does not require pre-training or joint optimization with conditional log-likelihood. Given a partial sequence generated by the model, we first identify the set of optimal suffixes that minimize the total edit distance, using an efficient dynamic programming algorithm. Then, for each position of the generated sequence, we use a target distribution which puts equal probability on the first token of all the optimal suffixes. OCD achieves the state-of-the-art performance on end-to-end speech recognition, on both Wall Street Journal and Librispeech datasets, achieving $9.3\\%$ WER and $4.5\\%$ WER, respectively.", "target": ["Optimal Completion Distillation (OCD) è una procedura di training per ottimizzare i modelli sequence-to-sequence basati sulla edit distance che raggiunge lo stato dell'arte nei task di riconoscimento vocale end-to-end.", "Approccio alternativo al training di modelli seq2seq utilizzando un programma dinamico per calcolare le prosecuzioni ottimali dei prefissi previsti", "Un algoritmo di training per modelli auto-regressivi che non richiede alcun pre-training MLE e può ottimizzare direttamente dal sampling.", "L'articolo considera una lacuna dei modelli sequence to sequence addestrati usando la stima di massima verosimiglianza e propone un approccio basato sulle edit distance e l'uso implicito di sequenze di label date durante il training."]} +{"source": "As an emerging field, federated learning has recently attracted considerable attention. Compared to distributed learning in the datacenter setting, federated learning has more strict constraints on computate efficiency of the learned model and communication cost during the training process. In this work, we propose an efficient federated learning framework based on variational dropout. Our approach is able to jointly learn a sparse model while reducing the amount of gradients exchanged during the iterative training process. We demonstrate the superior performance of our approach on achieving significant model compression and communication reduction ratios with no accuracy loss.", "target": ["Un modello congiunto e un metodo di sparsificazione del gradiente per federated learning", "Applica il dropout variazionale per ridurre il costo di comunicazione nel training distribuito delle reti neurali, e fa esperimenti sui dataset mnist, cifar10 e svhn.", "Gli autori propongono un algoritmo che riduce i costi di comunicazione nel federated learning inviando gradienti sparsi dal dispositivo al server e viceversa.", "Combina l'algoritmo di ottimizzazione distribuita con il dropout variazionale per sparsificare i gradienti inviati al server principale dai learner locali."]} +{"source": "We prove a multiclass boosting theory for the ResNet architectures which simultaneously creates a new technique for multiclass boosting and provides a new algorithm for ResNet-style architectures. Our proposed training algorithm, BoostResNet, is particularly suitable in non-differentiable architectures. Our method only requires the relatively inexpensive sequential training of T \"shallow ResNets\". We prove that the training error decays exponentially with the depth T if the weak module classifiers that we train perform slightly better than some weak baseline. In other words, we propose a weak learning condition and prove a boosting theory for ResNet under the weak learning condition. A generalization error bound based on margin theory is proved and suggests that ResNet could be resistant to overfitting using a network with l_1 norm bounded weights.", "target": ["Dimostriamo una teoria di boosting multiclasse per le architetture ResNet che crea contemporaneamente una nuova tecnica per il boosting multiclasse e fornisce un nuovo algoritmo per le architetture in stile ResNet.", "Presenta un algoritmo in stile boosting per il training di deep residual network, un'analisi di convergenza per l'errore di training e un'analisi della capacità di generalizzazione.", "Un metodo di apprendimento per ResNet utilizzando il framework di boosting che decompone l'apprendimento di reti complesse e prevede meno costi computazionali.", "Gli autori propongono la deep ResNet come algoritmo di boosting, e sostengono che questo è più efficiente della backpropagation standard end-to-end."]} +{"source": "We consider the problem of learning a one-hidden-layer neural network: we assume the input x is from Gaussian distribution and the label $y = a \\sigma(Bx) + \\xi$, where a is a nonnegative vector and $B$ is a full-rank weight matrix, and $\\xi$ is a noise vector. We first give an analytic formula for the population risk of the standard squared loss and demonstrate that it implicitly attempts to decompose a sequence of low-rank tensors simultaneously. \t Inspired by the formula, we design a non-convex objective function $G$ whose landscape is guaranteed to have the following properties:\t 1. All local minima of $G$ are also global minima. 2. All global minima of $G$ correspond to the ground truth parameters. 3. The value and gradient of $G$ can be estimated using samples. \t With these properties, stochastic gradient descent on $G$ provably converges to the global minimum and learn the ground-truth parameters. We also prove finite sample complexity results and validate the results by simulations.", "target": ["L'articolo analizza il panorama di ottimizzazione delle reti neurali con un hidden layer e progetta un nuovo obiettivo che non ha un minimo locale spurio.", "Questo articolo studia il problema dell'apprendimento delle reti neurali con un hidden layer, stabilisce una connessione tra la least square population loss e i polinomi di Hermite, e propone una nuova loss.", "Un metodo di fattorizzazione tensoriale per il learning di una rete neurale a un hidden layer"]} +{"source": "Open information extraction (OIE) systems extract relations and their arguments from natural language text in an unsupervised manner. The resulting extractions are a valuable resource for downstream tasks such as knowledge base construction, open question answering, or event schema induction. In this paper, we release, describe, and analyze an OIE corpus called OPIEC, which was extracted from the text of English Wikipedia. OPIEC complements the available OIE resources: It is the largest OIE corpus publicly available to date (over 340M triples) and contains valuable metadata such as provenance information, confidence scores, linguistic annotations, and semantic annotations including spatial and temporal information. We analyze the OPIEC corpus by comparing its content with knowledge bases such as DBpedia or YAGO, which are also based on Wikipedia. We found that most of the facts between entities present in OPIEC cannot be found in DBpedia and/or YAGO, that OIE facts often differ in the level of specificity compared to knowledge base facts, and that OIE open relations are generally highly polysemous. We believe that the OPIEC corpus is a valuable resource for future research on automated knowledge base construction.", "target": ["Un Open Information Extraction Corpus e la sua analisi approfondita", "Costruisce un nuovo corpus per information extraction che è più grande dei corpora pubblici precedenti e contiene informazioni non esistenti nei corpora attuali.", "Presenta un dataset di triple open-IE che sono state raccolte da Wikipedia con l'aiuto di un sistema di estrazione recente.", "L'articolo descrive la creazione di un corpus Open IE su Wikipedia inglese attraverso un modo automatico"]} +{"source": "The process of designing neural architectures requires expert knowledge and extensive trial and error. While automated architecture search may simplify these requirements, the recurrent neural network (RNN) architectures generated by existing methods are limited in both flexibility and components. We propose a domain-specific language (DSL) for use in automated architecture search which can produce novel RNNs of arbitrary depth and width. The DSL is flexible enough to define standard architectures such as the Gated Recurrent Unit and Long Short Term Memory and allows the introduction of non-standard RNN components such as trigonometric curves and layer normalization. Using two different candidate generation techniques, random search with a ranking function and reinforcement learning, we explore the novel architectures produced by the RNN DSL for language modeling and machine translation domains. The resulting architectures do not follow human intuition yet perform well on their targeted tasks, suggesting the space of usable RNN architectures is far larger than previously assumed.", "target": ["Definiamo un DSL flessibile per la generazione di architetture RNN che permette RNN di varie dimensioni e complessità e proponiamo una funzione di classificazione che rappresenta RNN come reti neurali ricorsive, simulando le loro prestazioni per decidere le architetture più promettenti.", "Introduce un nuovo metodo per generare architetture RNN utilizzando un linguaggio specifico del dominio per due tipi di generatori (casuale e basato su RL) insieme a una funzione di ranking e un valutatore.", "Questo articolo presenta la ricerca di buone architetture per RNN Cell come un problema di ottimizzazione black-box in cui gli esempi sono rappresentati come un albero di operatori e valutati sulla base di funzioni apprese o generate da un agente RL.", "Questo articolo indaga la strategia di meta-learning per la ricerca automatica dell'architettura nel contesto di RNN utilizzando un DSL che specifica le operazioni ricorrenti RNN."]} +{"source": "Researches on deep neural networks with discrete parameters and their deployment in embedded systems have been active and promising topics. Although previous works have successfully reduced precision in inference, transferring both training and inference processes to low-bitwidth integers has not been demonstrated simultaneously. In this work, we develop a new method termed as ``\"WAGE\" to discretize both training and inference, where weights (W), activations (A), gradients (G) and errors (E) among layers are shifted and linearly constrained to low-bitwidth integers. To perform pure discrete dataflow for fixed-point devices, we further replace batch normalization by a constant scaling layer and simplify other components that are arduous for integer implementation. Improved accuracies can be obtained on multiple datasets, which indicates that WAGE somehow acts as a type of regularization. Empirically, we demonstrate the potential to deploy training in hardware systems such as integer-based deep learning accelerators and neuromorphic chips with comparable accuracy and higher energy efficiency, which is crucial to future AI applications in variable scenarios with transfer and continual learning demands.", "target": ["Applichiamo il training e l'inferenza solo con numeri interi a bassa larghezza di bit nelle DNN", "Un metodo chiamato WAGE che quantizza tutti gli operandi e gli operatori in una rete neurale per ridurre il numero di bit di rappresentazione in una rete.", "Gli autori propongono pesi, attivazioni, gradienti ed errori discretizzati sia in fase di allenamento che di test sulle reti neurali"]} +{"source": "Modern Convolutional Neural Networks (CNNs) are complex, encompassing millions of parameters. Their deployment exerts computational, storage and energy demands, particularly on embedded platforms. Existing approaches to prune or sparsify CNNs require retraining to maintain inference accuracy. Such retraining is not feasible in some contexts. In this paper, we explore the sparsification of CNNs by proposing three model-independent methods. Our methods are applied on-the-fly and require no retraining. We show that the state-of-the-art models' weights can be reduced by up to 73% (compression factor of 3.7x) without incurring more than 5% loss in Top-5 accuracy. Additional fine-tuning gains only 8% in sparsity, which indicates that our fast on-the-fly methods are effective.", "target": ["In questo articolo, sviluppiamo metodi di sparsificazione veloci e privi di retraining che possono essere impiegati per la sparsificazione on-the-fly delle CNN in molti contesti industriali.", "Questo articolo propone approcci per il pruning delle CNN senza retraining, introducendo tre schemi per determinare le soglie sui pesi per il pruning.", "Questo articolo descrive un metodo per la sparsificazione delle CNN senza retraining."]} +{"source": "Curriculum learning and Self paced learning are popular topics in the machine learning that suggest to put the training samples in order by considering their difficulty levels. Studies in these topics show that starting with a small training set and adding new samples according to difficulty levels improves the learning performance. In this paper we experimented that we can also obtain good results by adding the samples randomly without a meaningful order. We compared our method with classical training, Curriculum learning, Self paced learning and their reverse ordered versions. Results of the statistical tests show that the proposed method is better than classical method and similar with the others. These results point a new training regime that removes the process of difficulty level determination in Curriculum and Self paced learning and as successful as these methods.", "target": ["Proponiamo che il training con set crescenti stage-by-stage fornisce un'ottimizzazione per le reti neurali.", "Gli autori paragonano il curriculum learning all'apprendimento in un ordine casuale con fasi che aggiungono un nuovo sample di esempi all'insieme precedentemente costruito in modo casuale", "Questo articolo studia l'influenza dell'ordinamento nel Curriculum e Self paced learning, e mostra che in una certa misura l'ordinamento delle istanze di training non è importante."]} +{"source": "We study the problem of learning to map, in an unsupervised way, between domains $A$ and $B$, such that the samples $\\vb \\in B$ contain all the information that exists in samples $\\va\\in A$ and some additional information. For example, ignoring occlusions, $B$ can be people with glasses, $A$ people without, and the glasses, would be the added information. When mapping a sample $\\va$ from the first domain to the other domain, the missing information is replicated from an independent reference sample $\\vb\\in B$. Thus, in the above example, we can create, for every person without glasses a version with the glasses observed in any face image. Our solution employs a single two-pathway encoder and a single decoder for both domains. The common part of the two domains and the separate part are encoded as two vectors, and the separate part is fixed at zero for domain $A$. The loss terms are minimal and involve reconstruction losses for the two domains and a domain confusion term. Our analysis shows that under mild assumptions, this architecture, which is much simpler than the literature guided-translation methods, is enough to ensure disentanglement between the two domains. We present convincing results in a few visual domains, such as no-glasses to glasses, adding facial hair based on a reference image, etc.", "target": ["Un metodo di traduzione image to image che aggiunge a un'immagine il contenuto di un'altra creando così una nuova immagine.", "Questo articolo affronta il task del transfer di contenuti, con una nuova loss."]} +{"source": "Mathematical reasoning---a core ability within human intelligence---presents some unique challenges as a domain: we do not come to understand and solve mathematical problems primarily on the back of experience and evidence, but on the basis of inferring, learning, and exploiting laws, axioms, and symbol manipulation rules. In this paper, we present a new challenge for the evaluation (and eventually the design) of neural architectures and similar system, developing a task suite of mathematics problems involving sequential questions and answers in a free-form textual input/output format. The structured nature of the mathematics domain, covering arithmetic, algebra, probability and calculus, enables the construction of training and test spits designed to clearly illuminate the capabilities and failure-modes of different architectures, as well as evaluate their ability to compose and relate knowledge and learned processes. Having described the data generation process and its potential future expansions, we conduct a comprehensive analysis of models from two broad classes of the most powerful sequence-to-sequence architectures and find notable differences in their ability to resolve mathematical problems and generalize their knowledge.", "target": ["Un dataset per testare il ragionamento matematico (e la generalizzazione algebrica), e i risultati sugli attuali modelli sequence-to-sequence.", "Presenta un nuovo dataset sintetico per valutare la capacità di ragionamento matematico dei modelli sequence-to-sequence, e lo usa per valutare diversi modelli.", "Modello per risolvere problemi matematici di base."]} +{"source": "Convolutional Neural Networks (CNNs) filter the input data using a series of spatial convolution operators with compactly supported stencils and point-wise nonlinearities. Commonly, the convolution operators couple features from all channels. For wide networks, this leads to immense computational cost in the training of and prediction with CNNs. In this paper, we present novel ways to parameterize the convolution more efficiently, aiming to decrease the number of parameters in CNNs and their computational complexity. We propose new architectures that use a sparser coupling between the channels and thereby reduce both the number of trainable weights and the computational cost of the CNN. Our architectures arise as new types of residual neural network (ResNet) that can be seen as discretizations of a Partial Differential Equations (PDEs) and thus have predictable theoretical properties. Our first architecture involves a convolution operator with a special sparsity structure, and is applicable to a large class of CNNs. Next, we present an architecture that can be seen as a discretization of a diffusion reaction PDE, and use it with three different convolution operators. We outline in our experiments that the proposed architectures, although considerably reducing the number of trainable weights, yield comparable accuracy to existing CNNs that are fully coupled in the channel dimension.", "target": ["Questo articolo introduce parametrizzazioni efficienti ed economiche di reti neurali convoluzionali motivate da equazioni differenziali parziali", "Introduce quattro alternative \"a basso costo\" all'operazione di convoluzione standard che possono essere usate al posto dell'operazione di convoluzione standard per ridurre la loro complessità computazionale.", "Questo articolo introduce metodi per ridurre il costo computazionale delle implementazioni CNN, e introduce nuove parametrizzazioni di CNN come architetture che limitano il parameter coupling.", "L'articolo propone una prospettiva basata su PDE per comprendere e parametrizzare le CNN"]} +{"source": "In this article we use rate-distortion theory, a branch of information theory devoted to the problem of lossy compression, to shed light on an important problem in latent variable modeling of data: is there room to improve the model? One way to address this question is to find an upper bound on the probability (equivalently a lower bound on the negative log likelihood) that the model can assign to some data as one varies the prior and/or the likelihood function in a latent variable model. The core of our contribution is to formally show that the problem of optimizing priors in latent variable models is exactly an instance of the variational optimization problem that information theorists solve when computing rate-distortion functions, and then to use this to derive a lower bound on negative log likelihood. Moreover, we will show that if changing the prior can improve the log likelihood, then there is a way to change the likelihood function instead and attain the same log likelihood, and thus rate-distortion theory is of relevance to both optimizing priors as well as optimizing likelihood functions. We will experimentally argue for the usefulness of quantities derived from rate-distortion theory in latent variable modeling by applying them to a problem in image modeling.", "target": ["Utilizzare la teoria della distorsione del tasso per dare un bound su quanto un modello con variabili latenti può essere migliorato", "Affronta i problemi di ottimizzazione del prior nel modello con variabili latenti e la selezione della funzione di verosimiglianza proponendo criteri basati su un limite inferiore della log-likelihood negativa.", "Presenta un teorema che dà un limite inferiore alla likelihood negativa del tasso di distorsione per la modellazione delle variabili latenti", "Gli autori sostengono che la teoria della distorsione del tasso per la compressione lossy fornisce uno strumento naturale per studiare i modelli con variabili latenti e propongono un lower bound."]} +{"source": "Backprop is the primary learning algorithm used in many machine learning algorithms. In practice, however, Backprop in deep neural networks is a highly sensitive learning algorithm and successful learning depends on numerous conditions and constraints. One set of constraints is to avoid weights that lead to saturated units. The motivation for avoiding unit saturation is that gradients vanish and as a result learning comes to a halt. Careful weight initialization and re-scaling schemes such as batch normalization ensure that input activity to the neuron is within the linear regime where gradients are not vanished and can flow. Here we investigate backpropagating error terms only linearly. That is, we ignore the saturation that arise by ensuring gradients always flow. We refer to this learning rule as Linear Backprop since in the backward pass the network appears to be linear. In addition to ensuring persistent gradient flow, Linear Backprop is also favorable when computation is expensive since gradients are never computed. Our early results suggest that learning with Linear Backprop is competitive with Backprop and saves expensive gradient computations.", "target": ["Ignoriamo le non linearità e non calcoliamo i gradienti nel backward pass per risparmiare calcoli e per assicurare che i gradienti scorrano sempre.", "L'autore ha proposto algoritmi di backprop lineare per garantire il flusso dei gradienti per tutte le parti durante la backpropagation."]} +{"source": "Deep neural networks with discrete latent variables offer the promise of better symbolic reasoning, and learning abstractions that are more useful to new tasks. There has been a surge in interest in discrete latent variable models, however, despite several recent improvements, the training of discrete latent variable models has remained challenging and their performance has mostly failed to match their continuous counterparts. Recent work on vector quantized autoencoders (VQ-VAE) has made substantial progress in this direction, with its perplexity almost matching that of a VAE on datasets such as CIFAR-10. In this work, we investigate an alternate training technique for VQ-VAE, inspired by its connection to the Expectation Maximization (EM) algorithm. Training the discrete autoencoder with EM and combining it with sequence level knowledge distillation alows us to develop a non-autoregressive machine translation model whose accuracy almost matches a strong greedy autoregressive baseline Transformer, while being 3.3 times faster at inference.", "target": ["Capire l'autoencoder discreto VQ-VAE sistematicamente usando EM e usarlo per progettare un modello di traduzione non autogressivo che corrisponde a una baseline autoregressiva forte.", "Questo articolo introduce un nuovo modo di interpretare il VQ-VAE e propone un nuovo algoritmo di training basato sul soft EM clustering.", "L'articolo presenta una visione alternativa sulla procedura di training per il VQ-VAE utilizzando l'algoritmo soft EM"]} +{"source": "Recent research about margin theory has proved that maximizing the minimum margin like support vector machines does not necessarily lead to better performance, and instead, it is crucial to optimize the margin distribution. In the meantime, margin theory has been used to explain the empirical success of deep network in recent studies. In this paper, we present ODN (the Optimal margin Distribution Network), a network which embeds a loss function in regard to the optimal margin distribution. We give a theoretical analysis for our method using the PAC-Bayesian framework, which confirms the significance of the margin distribution for classification within the framework of deep networks. In addition, empirical results show that the ODN model always outperforms the baseline cross-entropy loss model consistently across different regularization situations. And our ODN model also outperforms the cross-entropy loss (Xent), hinge loss and soft hinge loss model in generalization task through limited training data.", "target": ["Questo articolo presenta una deep neural network che incorpora una loss per quanto riguarda la distribuzione ottimale del margine, che allevia il problema dell'overfitting teoricamente ed empiricamente.", "Presenta un limite PAC-bayesiano per una margin loss "]} +{"source": "Deep network compression seeks to reduce the number of parameters in the network while maintaining a certain level of performance. Deep network distillation seeks to train a smaller network that matches soft-max performance of a larger network. While both regimes have led to impressive performance for their respective goals, neither provide insight into the importance of a given layer in the original model, which is useful if we are to improve our understanding of these highly parameterized models. In this paper, we present the concept of deep net triage, which individually assesses small blocks of convolution layers to understand their collective contribution to the overall performance, which we call \\emph{criticality}. We call it triage because we assess this criticality by answering the question: what is the impact to the health of the overall network if we compress a block of layers into a single layer. We propose a suite of triage methods and compare them on problem spaces of varying complexity. We ultimately show that, across these problem spaces, deep net triage is able to indicate the of relative importance of different layers. Surprisingly, our local structural compression technique also leads to an improvement in overall accuracy when the final model is fine-tuned globally.", "target": ["Cerchiamo di capire le rappresentazioni apprese nelle reti compresse attraverso un regime sperimentale che chiamiamo deep net triage", "Confronta vari metodi di inizializzazione e di training per trasferire la conoscenza dalla rete VGG a una student network più piccola, sostituendo blocchi di layer con layer singoli.", "Questo articolo presenta cinque metodi per fare il triage o la compressione dei blocchi di layer per le deep network.", "L'articolo propone un metodo per comprimere un blocco di layer in una NN che valuta diversi sotto-approcci"]} +{"source": "In this paper, we show a phenomenon, which we named ``super-convergence'', where residual networks can be trained using an order of magnitude fewer iterations than is used with standard training methods. The existence of super-convergence is relevant to understanding why deep networks generalize well. One of the key elements of super-convergence is training with cyclical learning rates and a large maximum learning rate. Furthermore, we present evidence that training with large learning rates improves performance by regularizing the network. In addition, we show that super-convergence provides a greater boost in performance relative to standard training when the amount of labeled training data is limited. We also derive a simplification of the Hessian Free optimization method to compute an estimate of the optimal learning rate. The architectures to replicate this work will be made available upon publication.", "target": ["La prova empirica di un nuovo fenomeno richiede nuove intuizioni teoriche ed è rilevante per le discussioni attive nella letteratura su SGD e la comprensione della generalizzazione.", "L'articolo discute un fenomeno per cui il training delle reti neurali in setting molto specifici può trarre molto profitto da una schedule che include grandi learning rate", "Gli autori analizzano il training delle residual network usando grandi learning rate ciclici, e dimostrano una convergenza veloce con i learning rate ciclici e la prova empirica che grandi learning rate posso fornire regolarizzazione."]} +{"source": "Infinite-width neural networks have been extensively used to study the theoretical properties underlying the extraordinary empirical success of standard, finite-width neural networks. Nevertheless, until now, infinite-width networks have been limited to at most two hidden layers. To address this shortcoming, we study the initialisation requirements of these networks and show that the main challenge for constructing them is defining the appropriate sampling distributions for the weights. Based on these observations, we propose a principled approach to weight initialisation that correctly accounts for the functional nature of the hidden layer activations and facilitates the construction of arbitrarily many infinite-width layers, thus enabling the construction of arbitrarily deep infinite-width networks. The main idea of our approach is to iteratively reparametrise the hidden-layer activations into appropriately defined reproducing kernel Hilbert spaces and use the canonical way of constructing probability distributions over these spaces for specifying the required weight distributions in a principled way. Furthermore, we examine the practical implications of this construction for standard, finite-width networks. In particular, we derive a novel weight initialisation scheme for standard, finite-width networks that takes into account the structure of the data and information about the task at hand. We demonstrate the effectiveness of this weight initialisation approach on the MNIST, CIFAR-10 and Year Prediction MSD datasets.", "target": ["Proponiamo un metodo per la costruzione di reti di larghezza infinita arbitrariamente profonde, sulla base del quale deriviamo un nuovo schema di inizializzazione dei pesi per le reti di larghezza finita e dimostriamo le sue prestazioni competitive.", "Propone un approccio di inizializzazione dei pesi per consentire reti infinitamente profonde e di larghezza infinita con risultati sperimentali su piccoli dataset.", "Propone deep neural network di larghezza infinita."]} +{"source": "Working memory requires information about external stimuli to be represented in the brain even after those stimuli go away. This information is encoded in the activities of neurons, and neural activities change over timescales of tens of milliseconds. Information in working memory, however, is retained for tens of seconds, suggesting the question of how time-varying neural activities maintain stable representations. Prior work shows that, if the neural dynamics are in the ` null space' of the representation - so that changes to neural activity do not affect the downstream read-out of stimulus information - then information can be retained for periods much longer than the time-scale of individual-neuronal activities. The prior work, however, requires precisely constructed synaptic connectivity matrices, without explaining how this would arise in a biological neural network. To identify mechanisms through which biological networks can self-organize to learn memory function, we derived biologically plausible synaptic plasticity rules that dynamically modify the connectivity matrix to enable information storing. Networks implementing this plasticity rule can successfully learn to form memory representations even if only 10% of the synapses are plastic, they are robust to synaptic noise, and they can represent information about multiple stimuli.", "target": ["Abbiamo derivato regole di apprendimento della synaptic plasticity biologicamente plausibili per una rete neurale ricorrente per memorizzare le rappresentazioni degli stimoli.", "Un modello di rete neurale composto da neuroni connessi in modo ricorrente e uno o più redout che mira a mantenere un certo output nel tempo.", "Questo articolo presenta un meccanismo di memoria auto-organizzante in un modello neurale, e introduce una funzione obiettivo che minimizza i cambiamenti nel segnale da memorizzare."]} +{"source": "Generative Adversarial Networks (GANs) have been proposed as an approach to learning generative models. While GANs have demonstrated promising performance on multiple vision tasks, their learning dynamics are not yet well understood, neither in theory nor in practice. In particular, the work in this domain has been focused so far only on understanding the properties of the stationary solutions that this dynamics might converge to, and of the behavior of that dynamics in this solutions’ immediate neighborhood. To address this issue, in this work we take a first step towards a principled study of the GAN dynamics itself. To this end, we propose a model that, on one hand, exhibits several of the common problematic convergence behaviors (e.g., vanishing gradient, mode collapse, diverging or oscillatory behavior), but on the other hand, is sufficiently simple to enable rigorous convergence analysis. This methodology enables us to exhibit an interesting phenomena: a GAN with an optimal discriminator provably converges, while guiding the GAN training using only a first order approximation of the discriminator leads to unstable GAN dynamics and mode collapse. This suggests that such usage of the first order approximation of the discriminator, which is a de-facto standard in all the existing GAN dynamics, might be one of the factors that makes GAN training so challenging in practice. Additionally, our convergence result constitutes the first rigorous analysis of a dynamics of a concrete parametric GAN.", "target": ["Per comprendere il training delle GAN, definiamo una semplice dinamica GAN e mostriamo le differenze quantitative tra gli aggiornamenti ottimali e quelli del primo ordine in questo modello.", "Gli autori studiano l'impatto delle GAN in setting in cui ad ogni iterazione il discriminatore si allena fino alla convergenza e il generatore si aggiorna con gradient step, o dove pochi gradient step sono fatti per il disciminatore e il generatore.", "Questo articolo studia la dinamica dell'adversarial training delle GAN su un modello a mistura gaussiana"]} +{"source": "The machine learning and computer vision community is witnessing an unprecedented rate of new tasks being proposed and addressed, thanks to the power of deep convolutional networks to find complex mappings from X to Y. The advent of each task often accompanies the release of a large-scale human-labeled dataset, for supervised training of the deep network. However, it is expensive and time-consuming to manually label sufficient amount of training data. Therefore, it is important to develop algorithms that can leverage off-the-shelf labeled dataset to learn useful knowledge for the target task. While previous works mostly focus on transfer learning from a single source, we study multi-source transfer across domains and tasks (MS-DTT), in a semi-supervised setting. We propose GradMix, a model-agnostic method applicable to any model trained with gradient-based learning rule. GradMix transfers knowledge via gradient descent, by weighting and mixing the gradients from all sources during training. Our method follows a meta-learning objective, by assigning layer-wise weights to the source gradients, such that the combined gradient follows the direction that can minimize the loss for a small set of samples from the target dataset. In addition, we propose to adaptively adjust the learning rate for each mini-batch based on its importance to the target task, and a pseudo-labeling method to leverage the unlabeled samples in the target domain. We perform experiments on two MS-DTT tasks: digit recognition and action recognition, and demonstrate the advantageous performance of the proposed method against multiple baselines.", "target": ["Proponiamo un metodo basato sul gradiente per trasferire la conoscenza da più fonti attraverso diversi domini e task.", "Questo articolo propone di combinare i gradienti dei domini di origine per aiutare l'apprendimento nel dominio target."]} +{"source": "Bayesian phylogenetic inference is currently done via Markov chain Monte Carlo with simple mechanisms for proposing new states, which hinders exploration efficiency and often requires long runs to deliver accurate posterior estimates. In this paper we present an alternative approach: a variational framework for Bayesian phylogenetic analysis. We approximate the true posterior using an expressive graphical model for tree distributions, called a subsplit Bayesian network, together with appropriate branch length distributions. We train the variational approximation via stochastic gradient ascent and adopt multi-sample based gradient estimators for different latent variables separately to handle the composite latent space of phylogenetic models. We show that our structured variational approximations are flexible enough to provide comparable posterior estimation to MCMC, while requiring less computation due to a more efficient tree exploration mechanism enabled by variational inference. Moreover, the variational approximations can be readily used for further statistical analysis such as marginal likelihood estimation for model comparison via importance sampling. Experiments on both synthetic data and real data Bayesian phylogenetic inference problems demonstrate the effectiveness and efficiency of our methods.", "target": ["La prima formulazione Bayes variazionale dell'inferenza filogenetica, un difficile problema di inferenza su strutture con componenti discrete e continue intrecciate", "Esplora una soluzione di inferenza approssimata al problema dell'inferenza bayesiana degli alberi filogenetici sfruttando le reti bayesiane subsplit recentemente proposte e i moderni stimatori di gradiente per VI.", "Propone un approccio variazionale all'inferenza posteriore bayesiana negli alberi filogenetici."]} +{"source": "This paper introduces HybridNet, a hybrid neural network to speed-up autoregressive models for raw audio waveform generation. As an example, we propose a hybrid model that combines an autoregressive network named WaveNet and a conventional LSTM model to address speech synthesis. Instead of generating one sample per time-step, the proposed HybridNet generates multiple samples per time-step by exploiting the long-term memory utilization property of LSTMs. In the evaluation, when applied to text-to-speech, HybridNet yields state-of-art performance. HybridNet achieves a 3.83 subjective 5-scale mean opinion score on US English, largely outperforming the same size WaveNet in terms of naturalness and provide 2x speed up at inference.", "target": ["È un'architettura neurale ibrida per accelerare il modello autoregressivo.", "Conclude che per scalare la dimensione del modello senza aumentare il tempo di inferenza per la predizione sequenziale si possa usare un modello che predice più timestep contemporaneamente.", "Questo articolo presenta HybridNet, un sistema di sintesi neurale del parlato e di altri audio che combina il modello WaveNet con una LSTM con l'obiettivo di offrire un modello con una generazione audio avente un tempo di inferenza più veloce."]} +{"source": "Visual Interpretation and explanation of deep models is critical towards wide adoption of systems that rely on them. In this paper, we propose a novel scheme for both interpretation as well as explanation in which, given a pretrained model, we automatically identify internal features relevant for the set of classes considered by the model, without relying on additional annotations. We interpret the model through average visualizations of this reduced set of features. Then, at test time, we explain the network prediction by accompanying the predicted class label with supporting visualizations derived from the identified features. In addition, we propose a method to address the artifacts introduced by strided operations in deconvNet-based visualizations. Moreover, we introduce an8Flower , a dataset specifically designed for objective quantitative evaluation of methods for visual explanation. Experiments on the MNIST , ILSVRC 12, Fashion 144k and an8Flower datasets show that our method produces detailed explanations with good coverage of relevant features of the classes of interest.", "target": ["Interpretazione identificando le feature apprese dal modello che servono come indicatori per il task di interesse. Si spiegano le decisioni del modello evidenziando la risposta di queste feature nei dati di test. Si valutano obiettivamente le spiegazioni con un dataset controllato.", "Questo articolo propone un metodo per produrre spiegazioni visive per gli output delle deep neural network e rilascia un nuovo dataset sintetico.", "Un metodo per deep neural network che identifica automaticamente le feature rilevanti dell'insieme delle classi, supportando l'interpretazione e la spiegazione senza fare affidamento su annotazioni aggiuntive."]} +{"source": "In this work we propose a simple and efficient framework for learning sentence representations from unlabelled data. Drawing inspiration from the distributional hypothesis and recent work on learning sentence representations, we reformulate the problem of predicting the context in which a sentence appears as a classification problem. Given a sentence and the context in which it appears, a classifier distinguishes context sentences from other contrastive sentences based on their vector representations. This allows us to efficiently learn different types of encoding functions, and we show that the model learns high-quality sentence representations. We demonstrate that our sentence representations outperform state-of-the-art unsupervised and supervised representation learning methods on several downstream NLP tasks that involve understanding sentence semantics while achieving an order of magnitude speedup in training time.", "target": ["Una framework per l'apprendimento efficiente di rappresentazioni di frasi di alta qualità.", "Propone un algoritmo più veloce per l'apprendimento di rappresentazioni di frasi in stile SkipThought da corpora di frasi ordinate che sostituisce il decoder a livello di parola con una contrastive loss di classificazione.", "Questo articolo propone un framework per l'apprendimento unsupervised di rappresentazioni di frasi massimizzando un modello di probabilità di frasi appartenenti al contesto rispetto a frasi candidate casuali"]} +{"source": "Many regularization methods have been proposed to prevent overfitting in neural networks. Recently, a regularization method has been proposed to optimize the variational lower bound of the Information Bottleneck Lagrangian. However, this method cannot be generalized to regular neural network architectures. We present the activation norm penalty that is derived from the information bottleneck principle and is theoretically grounded in a variation dropout framework. Unlike in previous literature, it can be applied to any general neural network. We demonstrate that this penalty can give consistent improvements to different state of the art architectures both in language modeling and image classification. We present analyses on the properties of this penalty and compare it to other methods that also reduce mutual information.", "target": ["Ricaviamo una norm penalty sull'output della rete neurale dal punto di vista dell'information bottleneck", "Propone Activation Norm Penalty, una regolarizzazione di tipo L_2 sulle attivazioni, derivandola dal principio di Information Bottleneck", "Questo articolo crea una mappatura tra le norm penalty di attivazione e il framework dell'information bottleneck usando il framework di dropout variazionale."]} +{"source": "Unsupervised learning of timeseries data is a challenging problem in machine learning. Here, we propose a novel algorithm, Deep Temporal Clustering (DTC), a fully unsupervised method, to naturally integrate dimensionality reduction and temporal clustering into a single end to end learning framework. The algorithm starts with an initial cluster estimates using an autoencoder for dimensionality reduction and a novel temporal clustering layer for cluster assignment. Then it jointly optimizes the clustering objective and the dimensionality reduction objective. Based on requirement and application, the temporal clustering layer can be customized with any temporal similarity metric. Several similarity metrics are considered and compared. To gain insight into features that the network has learned for its clustering, we apply a visualization method that generates a heat map of regions of interest in the timeseries. The viability of the algorithm is demonstrated using timeseries data from diverse domains, ranging from earthquakes to sensor data from spacecraft. In each case, we show that our algorithm outperforms traditional methods. This performance is attributed to fully integrated temporal dimensionality reduction and clustering criterion.", "target": ["Un metodo completamente unsupervised, per integrare naturalmente la riduzione della dimensionalità e il clustering temporale in un unico framework di apprendimento end-to-end.", "Propone un algoritmo che integra l'autoencoder con il clustering di dati di serie temporali usando una struttura di rete che si adatta ai dati di serie temporali.", "Un algoritmo per eseguire congiuntamente la riduzione della dimensionalità e il clustering temporale in un contesto di deep learning, utilizzando un autoencoder e un obiettivo di clustering.", "Gli autori hanno proposto un metodo di clustering unsupervised di serie temporali costruito con deep neural network e dotato di un encoder-decoder e una modalità di clustering per accorciare le serie temporali, estrarre le feature temporali locali e ottenere le rappresentazioni codificate."]} +{"source": "We study many-class few-shot (MCFS) problem in both supervised learning and meta-learning scenarios. Compared to the well-studied many-class many-shot and few-class few-shot problems, MCFS problem commonly occurs in practical applications but is rarely studied. MCFS brings new challenges because it needs to distinguish between many classes, but only a few samples per class are available for training. In this paper, we propose ``memory-augmented hierarchical-classification network (MahiNet)'' for MCFS learning. It addresses the ``many-class'' problem by exploring the class hierarchy, e.g., the coarse-class label that covers a subset of fine classes, which helps to narrow down the candidates for the fine class and is cheaper to obtain. MahiNet uses a convolutional neural network (CNN) to extract features, and integrates a memory-augmented attention module with a multi-layer perceptron (MLP) to produce the probabilities over coarse and fine classes. While the MLP extends the linear classifier, the attention module extends a KNN classifier, both together targeting the ''`few-shot'' problem. We design different training strategies of MahiNet for supervised learning and meta-learning. Moreover, we propose two novel benchmark datasets ''mcfsImageNet'' (as a subset of ImageNet) and ''mcfsOmniglot'' (re-splitted Omniglot) specifically for MCFS problem. In experiments, we show that MahiNet outperforms several state-of-the-art models on MCFS classification tasks in both supervised learning and meta-learning scenarios.", "target": ["Una rete neurale con memoria che affronta il problema many-class few-shot sfruttando la gerarchia delle classi sia nell'apprendimento supervisionato che nel meta-learning.", "Questo articolo presenta metodi per aggiungere un bias induttivo a un classificatore attraverso la predizione coarse-to-fine lungo una gerarchia di classi e l'apprendimento di un classificatore KNN memory-based che tiene traccia delle istanze mal annotate durante l'apprendimento.", "Questo articolo formula il problema di classificazione many-class few-shot da una prospettiva di apprendimento supervisionato e una prospettiva di meta-learning."]} +{"source": "Learning a better representation with neural networks is a challenging problem, which has been tackled from different perspectives in the past few years. In this work, we focus on learning a representation that would be useful in a clustering task. We introduce two novel loss components that substantially improve the quality of produced clusters, are simple to apply to arbitrary models and cost functions, and do not require a complicated training procedure. We perform an extensive set of experiments, supervised and unsupervised, and evaluate the proposed loss components on two most common types of models, Recurrent Neural Networks and Convolutional Neural Networks, showing that the approach we propose consistently improves the quality of KMeans clustering in terms of mutual information scores and outperforms previously proposed methods.", "target": ["Un nuovo componente di loss che costringe la rete ad apprendere una rappresentazione adatta al clustering durante l'allenamento per un task di classificazione.", "Questo articolo propone due termini di regolarizzazione basati su una hinge loss sulla divergenza KL tra due argomenti di input normalizzati con softmax per incoraggiare l'apprendimento di rappresentazioni disgiunte", "Proposta di due regolarizzatori destinati a rendere le rappresentazioni apprese nel penultimo layer di un classificatore più conformi alla struttura intrinseca dei dati."]} +{"source": "In high dimensions, the performance of nearest neighbor algorithms depends crucially on structure in the data. While traditional nearest neighbor datasets consisted mostly of hand-crafted feature vectors, an increasing number of datasets comes from representations learned with neural networks. We study the interaction between nearest neighbor algorithms and neural networks in more detail. We find that the network architecture can significantly influence the efficacy of nearest neighbor algorithms even when the classification accuracy is unchanged. Based on our experiments, we propose a number of training modifications that lead to significantly better datasets for nearest neighbor algorithms. Our modifications lead to learned representations that can accelerate nearest neighbor queries by 5x.", "target": ["Mostriamo come ottenere buone rappresentazioni dal punto di vista della Similarity Search.", "Studia l'impatto del cambiamento della parte di classificazione delle immagini a capo della DNN sulla capacità di indicizzare i descrittori con un LSH o un algoritmo kd-tree.", "Propone di utilizzare la softmax cross-entropy loss per imparare una rete che cerca di ridurre gli angoli tra gli input e i vettori di classe corrispondenti in un framework supervisionato."]} +{"source": "Neural network quantization has become an important research area due to its great impact on deployment of large models on resource constrained devices. In order to train networks that can be effectively discretized without loss of performance, we introduce a differentiable quantization procedure. Differentiability can be achieved by transforming continuous distributions over the weights and activations of the network to categorical distributions over the quantization grid. These are subsequently relaxed to continuous surrogates that can allow for efficient gradient-based optimization. We further show that stochastic rounding can be seen as a special case of the proposed approach and that under this formulation the quantization grid itself can also be optimized with gradient descent. We experimentally validate the performance of our method on MNIST, CIFAR 10 and Imagenet classification.", "target": ["Introduciamo una tecnica che permette il training basato sul gradiente delle reti neurali quantizzate.", "Propone un modo unificato e generale di addestrare reti neurali con pesi e attivazioni sinaptiche quantizzate a precisione ridotta.", "Un nuovo approccio alla quantizzazione delle attivazioni che è allo stato dell'arte o competitivo su diversi problemi di immagini reali.", "Un metodo per l'apprendimento di reti neurali con pesi e attivazioni quantizzate quantizzando stocasticamente i valori e sostituendo la distribuzione categoriale risultante con un rilassamento continuo della stessa"]} +{"source": "In most current formulations of adversarial training, the discriminators can be expressed as single-input operators, that is, the mapping they define is separable over observations. In this work, we argue that this property might help explain the infamous mode collapse phenomenon in adversarially-trained generative models. Inspired by discrepancy measures and two-sample tests between probability distributions, we propose distributional adversaries that operate on samples, i.e., on sets of multiple points drawn from a distribution, rather than on single observations. We show how they can be easily implemented on top of existing models. Various experimental results show that generators trained in combination with our distributional adversaries are much more stable and are remarkably less prone to mode collapse than traditional models trained with observation-wise prediction discriminators. In addition, the application of our framework to domain adaptation results in strong improvement over recent state-of-the-art.", "target": ["Mostriamo che il problema del mode collapse nelle GAN può essere spiegato da una mancanza di condivisione delle informazioni tra le osservazioni in un batch di training, e proponiamo un framework basato sulla distribuzione per la condivisione globale delle informazioni tra i gradienti che porta a un adversarial training più stabile ed efficace.", "Propone di sostituire i discriminatori a sample singolo nell'adversarial training con discriminatori che operano esplicitamente su distribuzioni di esempi.", "Teoria sui test a due sample e MMD e come possono essere beneficamente incorporati nel framework GAN."]} +{"source": "Chemical information extraction is to convert chemical knowledge in text into true chemical database, which is a text processing task heavily relying on chemical compound name identification and standardization. Once a systematic name for a chemical compound is given, it will naturally and much simply convert the name into the eventually required molecular formula. However, for many chemical substances, they have been shown in many other names besides their systematic names which poses a great challenge for this task. In this paper, we propose a framework to do the auto standardization from the non-systematic names to the corresponding systematic names by using the spelling error correction, byte pair encoding tokenization and neural sequence to sequence model. Our framework is trained end to end and is fully data-driven. Our standardization accuracy on the test dataset achieves 54.04% which has a great improvement compared to previous state-of-the-art result.", "target": ["Abbiamo progettato un framework end-to-end utilizzando il modello sequence to sequence per fare la standardizzazione dei nomi chimici.", "Standardizza i nomi non sistematici nella information extraction chimica creando un corpus parallelo di nomi non sistematici e sistematici e costruendo un modello seq2seq.", "Questo lavoro presenta un metodo per tradurre nomi non sistematici di composti chimici nei loro equivalenti sistematici utilizzando una combinazione di meccanismi"]} +{"source": "The training of deep neural networks with Stochastic Gradient Descent (SGD) with a large learning rate or a small batch-size typically ends in flat regions of the weight space, as indicated by small eigenvalues of the Hessian of the training loss. This was found to correlate with a good final generalization performance. In this paper we extend previous work by investigating the curvature of the loss surface along the whole training trajectory, rather than only at the endpoint. We find that initially SGD visits increasingly sharp regions, reaching a maximum sharpness determined by both the learning rate and the batch-size of SGD. At this peak value SGD starts to fail to minimize the loss along directions in the loss surface corresponding to the largest curvature (sharpest directions). To further investigate the effect of these dynamics in the training process, we study a variant of SGD using a reduced learning rate along the sharpest directions which we show can improve training speed while finding both sharper and better generalizing solution, compared to vanilla SGD. Overall, our results show that the SGD dynamics in the subspace of the sharpest directions influence the regions that SGD steers to (where larger learning rate or smaller batch size result in wider regions visited), the overall training speed, and the generalization ability of the final model.", "target": ["SGD viene indirizzato all'inizio dell'training verso una regione in cui il suo step è troppo grande rispetto alla curvatura, il che ha un impatto sul resto del training.", "Analizza la relazione tra la convergenza/generalizzazione e l'aggiornamento sui più grandi autovettori dell'Hessiana delle loss delle DNN.", "Questo lavoro studia la relazione tra la step size SGD e la curvatura della superficie di loss"]} +{"source": "We introduce a new approach to estimate continuous actions using actor-critic algorithms for reinforcement learning problems. Policy gradient methods usually predict one continuous action estimate or parameters of a presumed distribution (most commonly Gaussian) for any given state which might not be optimal as it may not capture the complete description of the target distribution. Our approach instead predicts M actions with the policy network (actor) and then uniformly sample one action during training as well as testing at each state. This allows the agent to learn a simple stochastic policy that has an easy to compute expected return. In all experiments, this facilitates better exploration of the state space during training and converges to a better policy.", "target": ["Introduciamo un nuovo algoritmo di reinforcement learning, che predice azioni multiple e fa sampling da esse.", "Questo lavoro introduce una miscela uniforme di policy deterministiche, e trova che questa parametrizzazione delle policy stocastiche supera DDPG su diversi benchmark della OpenAI gym.", "Gli autori studiano un metodo per migliorare le prestazioni delle reti addestrate con DDPG, e mostrano prestazioni migliorate su un gran numero di ambienti di controllo continuo standard."]} +{"source": "Recently convolutional neural networks (CNNs) achieve great accuracy in visual recognition tasks. DenseNet becomes one of the most popular CNN models due to its effectiveness in feature-reuse. However, like other CNN models, DenseNets also face overfitting problem if not severer. Existing dropout method can be applied but not as effective due to the introduced nonlinear connections. In particular, the property of feature-reuse in DenseNet will be impeded, and the dropout effect will be weakened by the spatial correlation inside feature maps. To address these problems, we craft the design of a specialized dropout method from three aspects, dropout location, dropout granularity, and dropout probability. The insights attained here could potentially be applied as a general approach for boosting the accuracy of other CNN models with similar nonlinear connections. Experimental results show that DenseNets with our specialized dropout method yield better accuracy compared to vanilla DenseNet and state-of-the-art CNN models, and such accuracy boost increases with the model depth.", "target": ["Rendendoci conto degli inconvenienti nell'applicazione del dropout originale su DenseNet, abbiamo realizzato il design di un metodo di dropout da tre aspetti, la cui idea potrebbe essere applicata anche ad altri modelli CNN.", "Applicazione di diverse strutture e schedule di dropout binario con l'obiettivo specifico di regolarizzare l'architettura DenseNet.", "Propone una tecnica di pre-dropout per densenet che implementa il dropout prima della funzione di attivazione non lineare."]} +{"source": "While extremely successful in several applications, especially with low-level representations; sparse, noisy samples and structured domains (with multiple objects and interactions) are some of the open challenges in most deep models. Column Networks, a deep architecture, can succinctly capture such domain structure and interactions, but may still be prone to sub-optimal learning from sparse and noisy samples. Inspired by the success of human-advice guided learning in AI, especially in data-scarce domains, we propose Knowledge-augmented Column Networks that leverage human advice/knowledge for better learning with noisy/sparse samples. Our experiments demonstrate how our approach leads to either superior overall performance or faster convergence.", "target": ["Guidare i deep model consapevoli delle relazioni verso un migliore apprendimento con la conoscenza umana.", "Questo lavoro propone una variante della rete a colonne basata sull'iniezione della guida umana modificando i calcoli nella rete.", "Un metodo per incorporare consigli umani nel deep learning estendendo Column Network, una graph neural network per la classificazione collettiva."]} +{"source": "Recent research has shown that one can train a neural network with binary weights and activations at train time by augmenting the weights with a high-precision continuous latent variable that accumulates small changes from stochastic gradient descent. However, there is a dearth of work to explain why one can effectively capture the features in data with binary weights and activations. Our main result is that the neural networks with binary weights and activations trained using the method of Courbariaux, Hubara et al. (2016) work because of the high-dimensional geometry of binary vectors. In particular, the ideal continuous vectors that extract out features in the intermediate representations of these BNNs are well-approximated by binary vectors in the sense that dot products are approximately preserved. Compared to previous research that demonstrated good classification performance with BNNs, our work explains why these BNNs work in terms of HD geometry. Furthermore, the results and analysis used on BNNs are shown to generalize to neural networks with ternary weights and activations. Our theory serves as a foundation for understanding not only BNNs but a variety of methods that seek to compress traditional neural networks. Furthermore, a better understanding of multilayer binary neural networks serves as a starting point for generalizing BNNs to other neural network architectures such as recurrent neural networks.", "target": ["I recenti successi delle Reti Neurali Binarie possono essere compresi sulla base della geometria dei vettori binari ad alta densità", "Indaga numericamente e teoricamente le ragioni dietro il successo empirico delle reti neurali binarizzate.", "Questo articolo analizza l'efficacia delle reti neurali binarie e perché la binarizzazione è in grado di preservare le prestazioni del modello."]} +{"source": "In recent years Convolutional Neural Networks (CNN) have been used extensively for Superresolution (SR). In this paper, we use inverse problem and sparse representation solutions to form a mathematical basis for CNN operations. We show how a single neuron is able to provide the optimum solution for inverse problem, given a low resolution image dictionary as an operator. Introducing a new concept called Representation Dictionary Duality, we show that CNN elements (filters) are trained to be representation vectors and then, during reconstruction, used as dictionaries. In the light of theoretical work, we propose a new algorithm which uses two networks with different structures that are separately trained with low and high coherency image patches and show that it performs faster compared to the state-of-the-art algorithms while not sacrificing from performance.", "target": ["Dopo aver dimostrato che un neurone agisce come un risolutore di problemi inversi per la super-risoluzione ed è garantito che una rete di neuroni sia in grado di fornire una soluzione, abbiamo proposto un'architettura di rete doppia che performa più velocemente dello stato dell'arte.", "Discute l'uso delle reti neurali per la super-risoluzione", "Una nuova architettura per risolvere task di super-risoluzione delle immagini, e un'analisi che mira a stabilire una connessione tra CNN per risolvere la super-risoluzione e risolvere problemi inversi regolarizzati sparsi."]} +{"source": "We consider the learning of algorithmic tasks by mere observation of input-output pairs. Rather than studying this as a black-box discrete regression problem with no assumption whatsoever on the input-output mapping, we concentrate on tasks that are amenable to the principle of divide and conquer, and study what are its implications in terms of learning. This principle creates a powerful inductive bias that we leverage with neural architectures that are defined recursively and dynamically, by learning two scale- invariant atomic operations: how to split a given input into smaller sets, and how to merge two partially solved tasks into a larger partial solution. Our model can be trained in weakly supervised environments, namely by just observing input-output pairs, and in even weaker environments, using a non-differentiable reward signal. Moreover, thanks to the dynamic aspect of our architecture, we can incorporate the computational complexity as a regularization term that can be optimized by backpropagation. We demonstrate the flexibility and efficiency of the Divide- and-Conquer Network on several combinatorial and geometric tasks: convex hull, clustering, knapsack and euclidean TSP. Thanks to the dynamic programming nature of our model, we show significant improvements in terms of generalization error and computational complexity.", "target": ["Modello dinamico che impara strategie di divide et impera tramite una supervisione debole.", "Propone di aggiungere un nuovo bias induttivo all'architettura delle reti neurali utilizzando una strategia divide et impera.", "Questo articolo studia i problemi che possono essere risolti usando un approccio di programmazione dinamica, e propone un'architettura di rete neurale per risolvere tali problemi che batte le baseline della sequenza.", "L'articolo propone un'architettura di rete unica che può imparare strategie divide et impera per risolvere task algoritmici."]} +{"source": "Within many machine learning algorithms, a fundamental problem concerns efficient calculation of an unbiased gradient wrt parameters $\\boldsymbol{\\gamma}$ for expectation-based objectives $\\mathbb{E}_{q_{\\boldsymbol{\\gamma}} (\\boldsymbol{y})} [f (\\boldsymbol{y}) ]$. Most existing methods either ($i$) suffer from high variance, seeking help from (often) complicated variance-reduction techniques; or ($ii$) they only apply to reparameterizable continuous random variables and employ a reparameterization trick. To address these limitations, we propose a General and One-sample (GO) gradient that ($i$) applies to many distributions associated with non-reparameterizable continuous {\\em or} discrete random variables, and ($ii$) has the same low-variance as the reparameterization trick. We find that the GO gradient often works well in practice based on only one Monte Carlo sample (although one can of course use more samples if desired). Alongside the GO gradient, we develop a means of propagating the chain rule through distributions, yielding statistical back-propagation, coupling neural networks to common random variables.", "target": ["Un gradiente Rep-like per distribuzioni continue/discrete non riparametrizzabili; ulteriormente generalizzato a deep probabilistic model, ottenendo una back-propagation statistica", "Presenta uno stimatore del gradiente per obiettivi basati sull'aspettazione che è unbiased, ha una bassa varianza e si applica a variabili casuali continue e discrete.", "Un metodo migliorato per calcolare le derivate dell'aspettazione e un nuovo stimatore di gradiente a bassa varianza che permette il training di modelli generativi in cui le osservazioni o le variabili latenti sono discrete.", "Formula un gradiente a bassa varianza per distribuzioni associate a variabili casuali continue o discrete."]} +{"source": "Quantum computers promise significant advantages over classical computers for a number of different applications. We show that the complete loss function landscape of a neural network can be represented as the quantum state output by a quantum computer. We demonstrate this explicitly for a binary neural network and, further, show how a quantum computer can train the network by manipulating this state using a well-known algorithm known as quantum amplitude amplification. We further show that with minor adaptation, this method can also represent the meta-loss landscape of a number of neural network architectures simultaneously. We search this meta-loss landscape with the same method to simultaneously train and design a binary neural network.", "target": ["Mostriamo che i parametri della NN e i landscape dei costi degli iperparametri possono essere generati come stati quantici usando un singolo circuito quantico e che questi possono essere usati per il training e il meta-training.", "Descrive un metodo in cui un framework di deep learning può essere quantizzato considerando la forma a due stati di una sfera di Bloch/qubit e creando una rete neurale binaria quantistica.", "Questo articolo propone l'amplificazione di ampiezza quantistica, un nuovo algoritmo per il training e la selezione del modello nelle reti neurali binarie.", "Propone un'idea nuova di produrre uno stato quantico che rappresenta un panorama completo dei costi di tutti i parametri per una data rete neurale binaria, costruendo una rete neurale binaria quantica (QBNN)."]} +{"source": "Several recent works have developed methods for training classifiers that are certifiably robust against norm-bounded adversarial perturbations. These methods assume that all the adversarial transformations are equally important, which is seldom the case in real-world applications. We advocate for cost-sensitive robustness as the criteria for measuring the classifier's performance for tasks where some adversarial transformation are more important than others. We encode the potential harm of each adversarial transformation in a cost matrix, and propose a general objective function to adapt the robust training method of Wong & Kolter (2018) to optimize for cost-sensitive robustness. Our experiments on simple MNIST and CIFAR10 models with a variety of cost matrices show that the proposed approach can produce models with substantially reduced cost-sensitive robust error, while maintaining classification accuracy.", "target": ["Un metodo generale per il training di classificatori robusti certificati sensibili ai costi contro le adversarial perturbation", "Calcola e inserisce i costi degli adversarial attack nell'obiettivo di ottimizzazione per ottenere un modello che sia robusto in termi di sensitività ai costi contro gli adversarial attack.", "Costruisce sul lavoro seminale di Dalvi et al. ed estende l'approccio alla robustezza certificabile con una matrice di costo che specifica per ogni coppia di classi source-target se il modello debba essere robusto agli adversarial example."]} +{"source": "Retinal prostheses for treating incurable blindness are designed to electrically stimulate surviving retinal neurons, causing them to send artificial visual signals to the brain. However, electrical stimulation generally cannot precisely reproduce normal patterns of neural activity in the retina. Therefore, an electrical stimulus must be selected that produces a neural response as close as possible to the desired response. This requires a technique for computing a distance between the desired response and the achievable response that is meaningful in terms of the visual signal being conveyed. Here we propose a method to learn such a metric on neural responses, directly from recorded light responses of a population of retinal ganglion cells (RGCs) in the primate retina. The learned metric produces a measure of similarity of RGC population responses that accurately reflects the similarity of the visual input. Using data from electrical stimulation experiments, we demonstrate that this metric may improve the performance of a prosthesis.", "target": ["Usare le triplette per imparare una metrica per confrontare le risposte neurali e migliorare le prestazioni di una protesi.", "Gli autori sviluppano nuove metriche di distanza degli spike train, comprese le reti neurali e le metriche quadratiche. Queste metriche hanno dimostrato di superare la metrica banale della distanza di Hamming, e catturano implicitamente alcune strutture nel codice neurale.", "Avendo in mente l'applicazione di migliorare le protesi neurali, gli autori propongono di imparare una metrica tra le risposte neurali ottimizzando una forma quadratica o una deep neural network."]} +{"source": "We introduce a novel workflow, QCue, for providing textual stimulation during mind-mapping. Mind-mapping is a powerful tool whose intent is to allow one to externalize ideas and their relationships surrounding a central problem. The key challenge in mind-mapping is the difficulty in balancing the exploration of different aspects of the problem (breadth) with a detailed exploration of each of those aspects (depth). Our idea behind QCue is based on two mechanisms: (1) computer-generated automatic cues to stimulate the user to explore the breadth of topics based on the temporal and topological evolution of a mind-map and (2) user-elicited queries for helping the user explore the depth for a given topic. We present a two-phase study wherein the first phase provided insights that led to the development of our work-flow for stimulating the user through cues and queries. In the second phase, we present a between-subjects evaluation comparing QCue with a digital mind-mapping work-flow without computer intervention. Finally, we present an expert rater evaluation of the mind-maps created by users in conjunction with user feedback.", "target": ["Questo articolo introduce un metodo per generare domande (spunti) e domande (suggerimenti) per aiutare gli utenti a eseguire il mind-mapping.", "Presenta uno strumento per assistere il mind-mapping attraverso il contesto suggerito relativo ai nodi esistenti e attraverso domande che espandono i rami meno sviluppati.", "Questo articolo presenta un approccio per assistere le persone nei task di mindmapping, progettando un'interfaccia e feature algoritmiche per supportare il mindmapping, e contribuisce a uno studio valutativo."]} +{"source": "The ability to detect when an input sample was not drawn from the training distribution is an important desirable property of deep neural networks. In this paper, we show that a simple ensembling of first and second order deep feature statistics can be exploited to effectively differentiate in-distribution and out-of-distribution samples. Specifically, we observe that the mean and standard deviation within feature maps differs greatly between in-distribution and out-of-distribution samples. Based on this observation, we propose a simple and efficient plug-and-play detection procedure that does not require re-training, pre-processing or changes to the model. The proposed method outperforms the state-of-the-art by a large margin in all standard benchmarking tasks, while being much simpler to implement and execute. Notably, our method improves the true negative rate from 39.6% to 95.3% when 95% of in-distribution (CIFAR-100) are correctly detected using a DenseNet and the out-of-distribution dataset is TinyImageNet resize. The source code of our method will be made publicly available.", "target": ["Rilevamento di sample fuori distribuzione utilizzando statistiche di feature di basso ordine senza richiedere alcun cambiamento nella DNN sottostante.", "Presenta un algoritmo per rilevare i sample fuori distribuzione utilizzando una running estimate della media e della varianza all'interno dei layer BatchNorm per costruire rappresentazioni di feature che vengono poi passate ad un classificatore lineare.", "Un approccio per il rilevamento di sample fuori distribuzione in cui gli autori propongono di utilizzare la regressione logistica sulle statistiche semplici di ogni layer di batch normalization della CNN.", "L'articolo suggerisce di usare i punteggi Z per confrontare i sample ID e OOD per valutare ciò che le deep network stanno cercando di fare."]} +{"source": "Due to the sharp increase in the severity of the threat imposed by software vulnerabilities, the detection of vulnerabilities in binary code has become an important concern in the software industry, such as the embedded systems industry, and in the field of computer security. However, most of the work in binary code vulnerability detection has relied on handcrafted features which are manually chosen by a select few, knowledgeable domain experts. In this paper, we attempt to alleviate this severe binary vulnerability detection bottleneck by leveraging recent advances in deep learning representations and propose the Maximal Divergence Sequential Auto-Encoder. In particular, latent codes representing vulnerable and non-vulnerable binaries are encouraged to be maximally divergent, while still being able to maintain crucial information from the original binaries. We conducted extensive experiments to compare and contrast our proposed methods with the baselines, and the results show that our proposed methods outperform the baselines in all performance measures of interest.", "target": ["Proponiamo un nuovo metodo chiamato Maximal Divergence Sequential Auto-Encoder che sfrutta la rappresentazione del Variational AutoEncoder per il rilevamento delle vulnerabilità del codice binario.", "Questo articolo propone un'architettura basata sull'autoencoder variazionale per l'embedding del codice per il rilevamento delle vulnerabilità del software binario, con il risultato che gli embedding appresi sono più efficaci nel distinguere tra codice binario vulnerabile e non vulnerabile rispetto alle baseline.", "Questo articolo propone un modello per estrarre automaticamente le feature per il rilevamento della vulnerabilità utilizzando la tecnica del deep learning."]} +{"source": "Modern neural architectures critically rely on attention for mapping structured inputs to sequences. In this paper we show that prevalent attention architectures do not adequately model the dependence among the attention and output tokens across a predicted sequence. We present an alternative architecture called Posterior Attention Models that after a principled factorization of the full joint distribution of the attention and output variables, proposes two major changes. First, the position where attention is marginalized is changed from the input to the output. Second, the attention propagated to the next decoding stage is a posterior attention distribution conditioned on the output. Empirically on five translation and two morphological inflection tasks the proposed posterior attention models yield better BLEU score and alignment accuracy than existing attention models.", "target": ["Calcolare l'attention in base alla distribuzione posteriore porta a un'attention più significativa e a prestazioni migliori", "Questo articolo propone un modello sequence-to-sequence in cui l'attention è trattata come una variabile latente, e deriva nuove procedure di inferenza per questo modello, ottenendo miglioramenti nella traduzione automatica e nei task di generazione di inflessioni morfologiche.", "Questo articolo presenta un nuovo modello di attention posteriore per problemi seq2seq"]} +{"source": "The growing interest to implement Deep Neural Networks (DNNs) on resource-bound hardware has motivated innovation of compression algorithms. Using these algorithms, DNN model sizes can be substantially reduced, with little to no accuracy degradation. This is achieved by either eliminating components from the model, or penalizing complexity during training. While both approaches demonstrate considerable compressions, the former often ignores the loss function during compression while the later produces unpredictable compressions. In this paper, we propose a technique that directly minimizes both the model complexity and the changes in the loss function. In this technique, we formulate compression as a constrained optimization problem, and then present a solution for it. We will show that using this technique, we can achieve competitive results.", "target": ["Comprimere i modelli DNN addestrati minimizzando la loro complessità mentre si limita la loro loss.", "Questo articolo propone un metodo per la compressione delle deep neural network sotto vincoli di precisione.", "Questo articolo presenta un metodo di codifica k-means vincolato al valore della loss per la compressione della rete e sviluppa un algoritmo iterativo per l'ottimizzazione del modello."]} +{"source": "Deep neural networks are able to solve tasks across a variety of domains and modalities of data. Despite many empirical successes, we lack the ability to clearly understand and interpret the learned mechanisms that contribute to such effective behaviors and more critically, failure modes. In this work, we present a general method for visualizing an arbitrary neural network's inner mechanisms and their power and limitations. Our dataset-centric method produces visualizations of how a trained network attends to components of its inputs. The computed \"attention masks\" support improved interpretability by highlighting which input attributes are critical in determining output. We demonstrate the effectiveness of our framework on a variety of deep neural network architectures in domains from computer vision and natural language processing. The primary contribution of our approach is an interpretable visualization of attention that provides unique insights into the network's underlying decision-making process irrespective of the data modality.", "target": ["Sviluppiamo una tecnica per visualizzare i meccanismi di attention in reti neurali arbitrarie.", "Propone di imparare una rete di attention latente che può aiutare a visualizzare la struttura interna di una deep neural network.", "Gli autori di questo articolo propongono uno schema di visualizzazione black-box guidato dai dati."]} +{"source": "The design of small molecules with bespoke properties is of central importance to drug discovery. However significant challenges yet remain for computational methods, despite recent advances such as deep recurrent networks and reinforcement learning strategies for sequence generation, and it can be difficult to compare results across different works. This work proposes 19 benchmarks selected by subject experts, expands smaller datasets previously used to approximately 1.1 million training molecules, and explores how to apply new reinforcement learning techniques effectively for molecular design. The benchmarks here, built as OpenAI Gym environments, will be open-sourced to encourage innovation in molecular design algorithms and to enable usage by those without a background in chemistry. Finally, this work explores recent development in reinforcement-learning methods with excellent sample complexity (the A2C and PPO algorithms) and investigates their behavior in molecular generation, demonstrating significant performance gains compared to standard reinforcement learning techniques.", "target": ["Indaghiamo una varietà di algoritmi RL per la generazione molecolare e definiamo nuovi benchmark (da rilasciare come OpenAI Gym), trovando che PPO e un algoritmo hill-climbing MLE funzionano meglio.", "Considera la valutazione del modello per la generazione di molecole proponendo 19 benchmark, espandendo piccoli dataset a un grande dataset standardizzato, ed esplorando come applicare le tecniche RL per la progettazione molecolare.", "Questo articolo mostra che i metodi RL più sofisticati sono meno efficaci della semplice tecnica hill-climbing (eccetto PPO) quando si modellano e sintetizzano le molecole."]} +{"source": "Analogical reasoning has been a principal focus of various waves of AI research. Analogy is particularly challenging for machines because it requires relational structures to be represented such that they can be flexibly applied across diverse domains of experience. Here, we study how analogical reasoning can be induced in neural networks that learn to perceive and reason about raw visual data. We find that the critical factor for inducing such a capacity is not an elaborate architecture, but rather, careful attention to the choice of data and the manner in which it is presented to the model. The most robust capacity for analogical reasoning is induced when networks learn analogies by contrasting abstract relational structures in their input domains, a training method that uses only the input data to force models to learn about important abstract features. Using this technique we demonstrate capacities for complex, visual and symbolic analogy making and generalisation in even the simplest neural network architectures.", "target": ["La capacità più robusta per il ragionamento analogico è indotta quando le reti imparano le analogie contrastando le strutture relazionali astratte nei loro domini di input.", "L'articolo studia la capacità di una rete neurale di imparare l'analogia, mostrando che una semplice rete neurale è in grado di risolvere alcuni problemi di analogia", "Questo articolo descrive un approccio per addestrare reti neurali per task di ragionamento analogico, considerando in particolare l'analogia visiva e le analogie simboliche."]} +{"source": "Building chatbots that can accomplish goals such as booking a flight ticket is an unsolved problem in natural language understanding. Much progress has been made to build conversation models using techniques such as sequence2sequence modeling. One challenge in applying such techniques to building goal-oriented conversation models is that maximum likelihood-based models are not optimized toward accomplishing goals. Recently, many methods have been proposed to address this issue by optimizing a reward that contains task status or outcome. However, adding the reward optimization on the fly usually provides little guidance for language construction and the conversation model soon becomes decoupled from the language model. In this paper, we propose a new setting in goal-oriented dialogue system to tighten the gap between these two aspects by enforcing model level information isolation on individual models between two agents. Language construction now becomes an important part in reward optimization since it is the only way information can be exchanged. We experimented our models using self-play and results showed that our method not only beat the baseline sequence2sequence model in rewards but can also generate human-readable meaningful conversations of comparable quality.", "target": ["Un modello di conversazione neurale goal-oriented attraverso il self-play", "Un modello di self-play per la generazione di dialoghi goal-oriented, che mira a rafforzare l'accoppiamento tra la reward del task e il language model.", "Questo articolo descrive un metodo per migliorare un sistema di dialogo goal-oriented usando il self-play."]} +{"source": "Search engine users nowadays heavily depend on query completion and correction to shape their queries. Typically, the completion is done by database lookup which does not understand the context and cannot generalize to prefixes not in the database . In the paper, we propose to use unsupervised deep language models to complete and correct the queries given an arbitrary prefix . We show how to address two main challenges that renders this method practical for large-scale deployment : 1) we propose a method for integrating error correction into the language model completion via a edit-distance potential and a variant of beam search that can exploit these potential functions; and 2) we show how to efficiently perform CPU-based computation to complete the queries, with error correction, in real time (generating top 10 completions within 16 ms). Experiments show that the method substantially increases hit rate over standard approaches, and is capable of handling tail queries.", "target": ["Completamento delle query di ricerca in tempo reale utilizzando language model LSTM a livello di carattere", "Questo articolo presenta metodi per il completamento delle query che includono la correzione dei prefissi, e alcuni dettagli tecnici per soddisfare particolari requisiti di latenza su una CPU.", "Gli autori propongono un algoritmo per risolvere il problema del completamento della query con la correzione degli errori, e adottano una modellazione basata su RNN a livello di carattere e ottimizzano la parte di inferenza per raggiungere gli obiettivi in tempo reale."]} +{"source": "RMSProp and ADAM continue to be extremely popular algorithms for training neural nets but their theoretical convergence properties have remained unclear. Further, recent work has seemed to suggest that these algorithms have worse generalization properties when compared to carefully tuned stochastic gradient descent or its momentum variants. In this work, we make progress towards a deeper understanding of ADAM and RMSProp in two ways. First, we provide proofs that these adaptive gradient algorithms are guaranteed to reach criticality for smooth non-convex objectives, and we give bounds on the running time. Next we design experiments to empirically study the convergence and generalization properties of RMSProp and ADAM against Nesterov's Accelerated Gradient method on a variety of common autoencoder setups and on VGG-9 with CIFAR-10. Through these experiments we demonstrate the interesting sensitivity that ADAM has to its momentum parameter \\beta_1. We show that at very high values of the momentum parameter (\\beta_1 = 0.99) ADAM outperforms a carefully tuned NAG on most of our experiments, in terms of getting lower training and test losses. On the other hand, NAG can sometimes do better when ADAM's \\beta_1 is set to the most commonly used value: \\beta_1 = 0.9, indicating the importance of tuning the hyperparameters of ADAM to get better generalization performance. We also report experiments on different autoencoders to demonstrate that NAG has better abilities in terms of reducing the gradient norms, and it also produces iterates which exhibit an increasing trend for the minimum eigenvalue of the Hessian of the loss function at the iterates.", "target": ["In questo articolo dimostriamo la convergenza alla criticità di RMSProp (stocastico e deterministico) e ADAM deterministico per obiettivi smooth non convessi e dimostriamo un'interessante sensibilità in termini di beta_1 per ADAM su autoencoder.", "Questo articolo presenta un'analisi di convergenza di RMSProp e ADAM nel caso di funzioni smooth non convesse"]} +{"source": "Recent advances in adversarial Deep Learning (DL) have opened up a new and largely unexplored surface for malicious attacks jeopardizing the integrity of autonomous DL systems. We introduce a novel automated countermeasure called Parallel Checkpointing Learners (PCL) to thwart the potential adversarial attacks and significantly improve the reliability (safety) of a victim DL model. The proposed PCL methodology is unsupervised, meaning that no adversarial sample is leveraged to build/train parallel checkpointing learners. We formalize the goal of preventing adversarial attacks as an optimization problem to minimize the rarely observed regions in the latent feature space spanned by a DL network. To solve the aforementioned minimization problem, a set of complementary but disjoint checkpointing modules are trained and leveraged to validate the victim model execution in parallel. Each checkpointing learner explicitly characterizes the geometry of the input data and the corresponding high-level data abstractions within a particular DL layer. As such, the adversary is required to simultaneously deceive all the defender modules in order to succeed. We extensively evaluate the performance of the PCL methodology against the state-of-the-art attack scenarios, including Fast-Gradient-Sign (FGS), Jacobian Saliency Map Attack (JSMA), Deepfool, and Carlini&WagnerL2 algorithm. Extensive proof-of-concept evaluations for analyzing various data collections including MNIST, CIFAR10, and ImageNet corroborate the effectiveness of our proposed defense mechanism against adversarial samples.", "target": ["L'elaborazione di meccanismi di difesa non supervisionati contro gli adversarial attack è fondamentale per garantire la generalizzabilità della difesa.", "Questo articolo presenta un metodo per rilevare adversarial example in un ambiente di classificazione di deep learning", "Questo articolo presenta un metodo unsupervised per rilevare adversarial example di reti neurali."]} +{"source": "Neural architecture search (NAS) has a great impact by automatically designing effective neural network architectures. However, the prohibitive computational demand of conventional NAS algorithms (e.g. 10 4 GPU hours) makes it difficult to directly search the architectures on large-scale tasks (e.g. ImageNet). Differentiable NAS can reduce the cost of GPU hours via a continuous representation of network architecture but suffers from the high GPU memory consumption issue (grow linearly w.r.t. candidate set size). As a result, they need to utilize proxy tasks, such as training on a smaller dataset, or learning with only a few blocks, or training just for a few epochs. These architectures optimized on proxy tasks are not guaranteed to be optimal on the target task. In this paper, we present ProxylessNAS that can directly learn the architectures for large-scale target tasks and target hardware platforms. We address the high memory consumption issue of differentiable NAS and reduce the computational cost (GPU hours and GPU memory) to the same level of regular training while still allowing a large candidate set. Experiments on CIFAR-10 and ImageNet demonstrate the effectiveness of directness and specialization. On CIFAR-10, our model achieves 2.08% test error with only 5.7M parameters, better than the previous state-of-the-art architecture AmoebaNet-B, while using 6× fewer parameters. On ImageNet, our model achieves 3.1% better top-1 accuracy than MobileNetV2, while being 1.2× faster with measured GPU latency. We also apply ProxylessNAS to specialize neural architectures for hardware with direct hardware metrics (e.g. latency) and provide insights for efficient CNN architecture design.", "target": ["Ricerca di architetture neurali senza proxy per l'apprendimento diretto di architetture su task di larga scala (ImageNet) riducendo il costo allo stesso livello del normale training.", "Questo articolo affronta il problema della ricerca dell'architettura, e in particolare cerca di farlo senza doversi allenare su task \"proxy\" dove il problema è semplificato per via di un'ottimizzazione, di una complessità dell'architettura o di una dimensione del dataset più trattabili."]} +{"source": "With the recently rapid development in deep learning, deep neural networks have been widely adopted in many real-life applications. However, deep neural networks are also known to have very little control over its uncertainty for test examples, which potentially causes very harmful and annoying consequences in practical scenarios. In this paper, we are particularly interested in designing a higher-order uncertainty metric for deep neural networks and investigate its performance on the out-of-distribution detection task proposed by~\\cite{hendrycks2016baseline}. Our method first assumes there exists a underlying higher-order distribution $\\mathcal{P}(z)$ , which generated label-wise distribution $\\mathcal{P}(y)$ over classes on the K-dimension simplex, and then approximate such higher-order distribution via parameterized posterior function $p_{\\theta}(z|x)$ under variational inference framework, finally we use the entropy of learned posterior distribution $p_{\\theta}(z|x)$ as uncertainty measure to detect out-of-distribution examples. However , we identify the overwhelming over-concentration issue in such a framework, which greatly hinders the detection performance. Therefore , we further design a log-smoothing function to alleviate such issue to greatly increase the robustness of the proposed entropy-based uncertainty measure. Through comprehensive experiments on various datasets and architectures, our proposed variational Dirichlet framework with entropy-based uncertainty measure is consistently observed to yield significant improvements over many baseline systems.", "target": ["Un nuovo framework basato sull'inferenza variazionale per out-of-distribution detection", "Descrive un approccio probabilistico per quantificare l'incertezza nei task di classificazione DNN che supera altri metodi SOTA nel task di out-of-distribution detection.", "Un nuovo framework per il rilevamento di out-of-distribution, basato sull'inferenza variazionale e una distribuzione di Dirichlet a priori, che riporta risultati allo stato dell'arte su diversi dataset.", "Out-of-distribution detection tramite un nuovo metodo per approssimare la distribuzione di fiducia della probabilità di classificazione usando l'inferenza variazionale della distribuzione di Dirichlet."]} +{"source": "Intelligent agents can learn to represent the action spaces of other agents simply by observing them act. Such representations help agents quickly learn to predict the effects of their own actions on the environment and to plan complex action sequences. In this work, we address the problem of learning an agent’s action space purely from visual observation. We use stochastic video prediction to learn a latent variable that captures the scene's dynamics while being minimally sensitive to the scene's static content. We introduce a loss term that encourages the network to capture the composability of visual sequences and show that it leads to representations that disentangle the structure of actions. We call the full model with composable action representations Composable Learned Action Space Predictor (CLASP). We show the applicability of our method to synthetic settings and its potential to capture action spaces in complex, realistic visual settings. When used in a semi-supervised setting, our learned representations perform comparably to existing fully supervised methods on tasks such as action-conditioned video prediction and planning in the learned action space, while requiring orders of magnitude fewer action labels. Project website: https://daniilidis-group.github.io/learned_action_spaces", "target": ["Impariamo una rappresentazione dello spazio d'azione di un agente dalle pure osservazioni visive. Usiamo un approccio con variabili latenti ricorrenti con una nuova loss di componibilità.", "Propone un modello compositivo a variabili latenti per imparare modelli che prevedano cosa accadrà dopo in scenari in cui le action-label non sono disponibili in abbondanza.", "Un approccio basato su IB variazionale per imparare le rappresentazioni delle azioni direttamente dai video delle azioni compiute, ottenendo una migliore efficienza dei metodi di apprendimento successivi e richiedendo una minore quantità di video di action label.", "Questo articolo propone un approccio alla predizione video che trova autonomamente uno spazio d'azione che codifica le differenze tra i fotogrammi successivi"]} +{"source": "When autonomous agents interact in the same environment, they must often cooperate to achieve their goals. One way for agents to cooperate effectively is to form a team, make a binding agreement on a joint plan, and execute it. However, when agents are self-interested, the gains from team formation must be allocated appropriately to incentivize agreement. Various approaches for multi-agent negotiation have been proposed, but typically only work for particular negotiation protocols. More general methods usually require human input or domain-specific data, and so do not scale. To address this, we propose a framework for training agents to negotiate and form teams using deep reinforcement learning. Importantly, our method makes no assumptions about the specific negotiation protocol, and is instead completely experience driven. We evaluate our approach on both non-spatial and spatially extended team-formation negotiation environments, demonstrating that our agents beat hand-crafted bots and reach negotiation outcomes consistent with fair solutions predicted by cooperative game theory. Additionally, we investigate how the physical location of agents influences negotiation outcomes.", "target": ["Il reinforcement learning può essere usato per addestrare gli agenti a negoziare la formazione della squadra attraverso molti protocolli di negoziazione", "Questo articolo studia il deep RL multi-agente in ambienti in cui tutti gli agenti devono cooperare per portare a termine un task (ad esempio, ricerca e salvataggio, videogiochi multi-player), e utilizza semplici giochi di voto ponderato cooperativo per studiare l'efficacia del deep RL e per confrontare le soluzioni trovate dal deep RL con una soluzione fair.", "Un approccio di reinforcement learning per la negoziazione di coalizioni in setting di teoria dei giochi cooperativi che può essere utilizzato nei casi in cui sono disponibili simulazioni di allenamento illimitate."]} +{"source": "Neural machine translation (NMT) models learn representations containing substantial linguistic information. However, it is not clear if such information is fully distributed or if some of it can be attributed to individual neurons. We develop unsupervised methods for discovering important neurons in NMT models. Our methods rely on the intuition that different models learn similar properties, and do not require any costly external supervision. We show experimentally that translation quality depends on the discovered neurons, and find that many of them capture common linguistic phenomena. Finally, we show how to control NMT translations in predictable ways, by modifying activations of individual neurons.", "target": ["Metodi non supervisionati per trovare, analizzare e controllare i neuroni importanti nella NMT", "Questo articolo presenta approcci non supervisionati per scoprire i neuroni importanti nei sistemi di traduzione automatica neurale e analizza le proprietà linguistiche controllate da questi neuroni.", "Metodi non supervisionati per classificare i neuroni nella traduzione automatica, dove i neuroni importanti sono così identificati e utilizzati per controllare l'output della traduzione automatica."]} +{"source": "Recent state-of-the-art reinforcement learning algorithms are trained under the goal of excelling in one specific task. Hence, both environment and task specific knowledge are entangled into one framework. However, there are often scenarios where the environment (e.g. the physical world) is fixed while only the target task changes. Hence, borrowing the idea from hierarchical reinforcement learning, we propose a framework that disentangles task and environment specific knowledge by separating them into two units. The environment-specific unit handles how to move from one state to the target state; and the task-specific unit plans for the next target state given a specific task. The extensive results in simulators indicate that our method can efficiently separate and learn two independent units, and also adapt to a new task more efficiently than the state-of-the-art methods.", "target": ["Proponiamo un framework DRL che distingue la conoscenza specifica del task e dell'ambiente.", "Gli autori propongono di decomporre reinforcement learning in una funzione PATH e una funzione GOAL", "Un'architettura modulare con l'obiettivo di separare la conoscenza specifica dell'ambiente e la conoscenza specifica del task in diversi moduli, alla pari con l'A3C standard in una vasta gamma di task."]} +{"source": "Modelling 3D scenes from 2D images is a long-standing problem in computer vision with implications in, e.g., simulation and robotics. We propose pix2scene, a deep generative-based approach that implicitly models the geometric properties of a scene from images. Our method learns the depth and orientation of scene points visible in images. Our model can then predict the structure of a scene from various, previously unseen view points. It relies on a bi-directional adversarial learning mechanism to generate scene representations from a latent code, inferring the 3D representation of the underlying scene geometry. We showcase a novel differentiable renderer to train the 3D model in an end-to-end fashion, using only images. We demonstrate the generative ability of our model qualitatively on both a custom dataset and on ShapeNet. Finally, we evaluate the effectiveness of the learned 3D scene representation in supporting a 3D spatial reasoning.", "target": ["pix2scene: un approccio generativo profondo per modellare implicitamente le proprietà geometriche di una scena 3D a partire dalle immagini", "Esplora la spiegazione di scene con surfel in un modello di riconoscimento neurale, e dimostra i risultati sulla ricostruzione e la sintesi dell'immagine e mental shape rotation.", "Gli autori introducono un metodo per creare un modello di scena 3D dato un'immagine 2D e una posa della telecamera utilizzando un modello self-supervised"]} +{"source": "Identifying the relations that connect words is an important step towards understanding human languages and is useful for various NLP tasks such as knowledge base completion and analogical reasoning. Simple unsupervised operators such as vector offset between two-word embeddings have shown to recover some specific relationships between those words, if any. Despite this, how to accurately learn generic relation representations from word representations remains unclear. We model relation representation as a supervised learning problem and learn parametrised operators that map pre-trained word embeddings to relation representations. We propose a method for learning relation representations using a feed-forward neural network that performs relation prediction. Our evaluations on two benchmark datasets reveal that the penultimate layer of the trained neural network-based relational predictor acts as a good representation for the relations between words.", "target": ["Identificare le relazioni che collegano le parole è importante per vari task NLP. Modelliamo la rappresentazione delle relazioni come un problema di apprendimento supervisionato e impariamo operatori parametrici che mappano i word embedding pre-addestrati alle rappresentazioni delle relazioni.", "Questo articolo presenta un nuovo metodo per rappresentare le relazioni lessicali come vettori usando solo i word embedding pre-addestrati e una nuova loss che opera su coppie di coppie di parole.", "Una nuova soluzione al problema della composizione delle relazioni quando si hanno già embedding di parole/entità preaddestrati e si è interessati solo ad imparare a comporre rappresentazioni di relazioni."]} +{"source": "Recurrent neural networks (RNNs) are important class of architectures among neural networks useful for language modeling and sequential prediction. However, optimizing RNNs is known to be harder compared to feed-forward neural networks. A number of techniques have been proposed in literature to address this problem. In this paper we propose a simple technique called fraternal dropout that takes advantage of dropout to achieve this goal. Specifically, we propose to train two identical copies of an RNN (that share parameters) with different dropout masks while minimizing the difference between their (pre-softmax) predictions. In this way our regularization encourages the representations of RNNs to be invariant to dropout mask, thus being robust. We show that our regularization term is upper bounded by the expectation-linear dropout objective which has been shown to address the gap due to the difference between the train and inference phases of dropout. We evaluate our model and achieve state-of-the-art results in sequence modeling tasks on two benchmark datasets - Penn Treebank and Wikitext-2. We also show that our approach leads to performance improvement by a significant margin in image captioning (Microsoft COCO) and semi-supervised (CIFAR-10) tasks.", "target": ["Proponiamo di addestrare due copie identiche di una rete neurale ricorrente (che condividono i parametri) con diverse maschere di dropout, minimizzando la differenza tra le loro predizioni (pre-softmax).", "Presenta Fraternal dropout come un miglioramento rispetto a Expectation-linear dropout in termini di convergenza, e dimostra l'utilità di Fraternal dropout su una serie di task e dataset."]} +{"source": "We propose a novel approach for deformation-aware neural networks that learn the weighting and synthesis of dense volumetric deformation fields. Our method specifically targets the space-time representation of physical surfaces from liquid simulations. Liquids exhibit highly complex, non-linear behavior under changing simulation conditions such as different initial conditions. Our algorithm captures these complex phenomena in two stages: a first neural network computes a weighting function for a set of pre-computed deformations, while a second network directly generates a deformation field for refining the surface. Key for successful training runs in this setting is a suitable loss function that encodes the effect of the deformations, and a robust calculation of the corresponding gradients. To demonstrate the effectiveness of our approach, we showcase our method with several complex examples of flowing liquids with topology changes. Our representation makes it possible to rapidly generate the desired implicit surfaces. We have implemented a mobile application to demonstrate that real-time interactions with complex liquid effects are possible with our approach.", "target": ["Apprendimento di ponderazioni e deformazioni di dataset spazio-temporali per approssimazioni altamente efficienti del comportamento dei liquidi.", "Un modello basato su reti neurali è usato per interpolare simulazioni per nuove condizioni di scena da superfici implicite 4D densamente registrate per una scena strutturata.", "Questo articolo presenta un approccio di deep learning accoppiato per generare dati realistici di simulazione di liquidi che possono essere utili per applicazioni di supporto decisionale in tempo reale.", "Questo articolo introduce un approccio di deep learning per la simulazione fisica che combina due reti per sintetizzare dati 4D che rappresentano simulazioni fisiche 3D"]} +{"source": "This is an empirical paper which constructs color invariant networks and evaluates their performances on a realistic data set. The paper studies the simplest possible case of color invariance: invariance under pixel-wise permutation of the color channels. Thus the network is aware not of the specific color object, but its colorfulness. The data set introduced in the paper consists of images showing crashed cars from which ten classes were extracted. An additional annotation was done which labeled whether the car shown was red or non-red. The networks were evaluated by their performance on the classification task. With the color annotation we altered the color ratios in the training data and analyzed the generalization capabilities of the networks on the unaltered test data. We further split the test data in red and non-red cars and did a similar evaluation. It is shown in the paper that an pixel-wise ordering of the rgb-values of the images performs better or at least similarly for small deviations from the true color ratios. The limits of these networks are also discussed.", "target": ["Costruiamo e valutiamo reti neurali invarianti al colore su un nuovo dataset realistico", "Propone un metodo per rendere invarianti al colore le reti neurali per il riconoscimento delle immagini e lo valuta sul dataset cifar 10.", "Gli autori studiano un layer di input modificato che risulta in reti invarianti al colore, e mostrano che certi layer di input invarianti al colore possono migliorare l'accuratezza per le immagini di test provenienti da una distribuzione di colore diversa dalle immagini di training.", "Gli autori testano una CNN su immagini con canali di colore modificati per essere invarianti alle permutazioni, con prestazioni non troppo degradate."]} +{"source": "Expressive efficiency refers to the relation between two architectures A and B, whereby any function realized by B could be replicated by A, but there exists functions realized by A, which cannot be replicated by B unless its size grows significantly larger. For example, it is known that deep networks are exponentially efficient with respect to shallow networks, in the sense that a shallow network must grow exponentially large in order to approximate the functions represented by a deep network of polynomial size. In this work, we extend the study of expressive efficiency to the attribute of network connectivity and in particular to the effect of \"overlaps\" in the convolutional process, i.e., when the stride of the convolution is smaller than its filter size (receptive field). To theoretically analyze this aspect of network's design, we focus on a well-established surrogate for ConvNets called Convolutional Arithmetic Circuits (ConvACs), and then demonstrate empirically that our results hold for standard ConvNets as well. Specifically, our analysis shows that having overlapping local receptive fields, and more broadly denser connectivity, results in an exponential increase in the expressive capacity of neural networks. Moreover, while denser connectivity can increase the expressive capacity, we show that the most common types of modern architectures already exhibit exponential increase in expressivity, without relying on fully-connected layers.", "target": ["Analizziamo come il grado di sovrapposizione tra i campi recettivi di una rete convoluzionale influenza la sua potenza espressiva.", "L'articolo studia la potenza espressiva fornita dalla \"sovrapposizione\" nei layer di convoluzione delle DNN considerando attivazioni lineari con il product pooling.", "Questo articolo analizza l'espressività dei circuiti aritmetici convoluzionali e mostra che un numero esponenzialmente grande di ConvAC non sovrapposti è richiesto per approssimare il grid tensor di un ConvAC sovrapposto."]} +{"source": "We provide a theoretical algorithm for checking local optimality and escaping saddles at nondifferentiable points of empirical risks of two-layer ReLU networks. Our algorithm receives any parameter value and returns: local minimum, second-order stationary point, or a strict descent direction. The presence of M data points on the nondifferentiability of the ReLU divides the parameter space into at most 2^M regions, which makes analysis difficult. By exploiting polyhedral geometry, we reduce the total computation down to one convex quadratic program (QP) for each hidden node, O(M) (in)equality tests, and one (or a few) nonconvex QP. For the last QP, we show that our specific problem can be solved efficiently, in spite of nonconvexity. In the benign case, we solve one equality constrained QP, and we prove that projected gradient descent solves it exponentially fast. In the bad case, we have to solve a few more inequality constrained QPs, but we prove that the time complexity is exponential only in the number of inequality constraints. Our experiments show that either benign case or bad case with very few inequality constraints occurs, implying that our algorithm is efficient in most cases.", "target": ["Un algoritmo teorico per testare l'ottimalità locale ed estrarre le direzioni di discesa in punti non differenziabili di rischi empirici di reti ReLU a un hidden layer.", "Propone un algoritmo per verificare se un dato punto è un punto stazionario generalizzato del secondo ordine.", "Un algoritmo teorico, che coinvolge la risoluzione di programmi quadratici convessi e non convessi, per verificare l'ottimalità locale e sfuggire ai punti di sella quando si addestrano reti ReLU a due layer.", "L'autore propone un metodo per controllare se un punto è un punto stazionario o no e poi classificare i punti stazionari come min locali o stazionari di secondo ordine"]} +{"source": "We present a new technique for learning visual-semantic embeddings for cross-modal retrieval. Inspired by the use of hard negatives in structured prediction, and ranking loss functions used in retrieval, we introduce a simple change to common loss functions used to learn multi-modal embeddings. That, combined with fine-tuning and the use of augmented data, yields significant gains in retrieval performance. We showcase our approach, dubbed VSE++, on the MS-COCO and Flickr30K datasets, using ablation studies and comparisons with existing methods. On MS-COCO our approach outperforms state-of-the-art methods by 8.8% in caption retrieval, and 11.3% in image retrieval (based on R@1).", "target": ["Una nuova loss basata su negativi relativamente hard che raggiunge prestazioni allo stato dell'arte nel recupero delle image-caption.", "Imparare l'embedding congiunto di frasi e immagini usando la triplet loss che viene applicata ai negativi più difficili invece di fare la media su tutte le triplet"]} +{"source": "We present DANTE, a novel method for training neural networks, in particular autoencoders, using the alternating minimization principle. DANTE provides a distinct perspective in lieu of traditional gradient-based backpropagation techniques commonly used to train deep networks. It utilizes an adaptation of quasi-convex optimization techniques to cast autoencoder training as a bi-quasi-convex optimization problem. We show that for autoencoder configurations with both differentiable (e.g. sigmoid) and non-differentiable (e.g. ReLU) activation functions, we can perform the alternations very effectively. DANTE effortlessly extends to networks with multiple hidden layers and varying network configurations. In experiments on standard datasets, autoencoders trained using the proposed method were found to be very promising when compared to those trained using traditional backpropagation techniques, both in terms of training speed, as well as feature extraction and reconstruction performance.", "target": ["Utilizziamo il principio di minimizzazione alternata per fornire una nuova tecnica efficace per addestrare i deep autoencoder.", "Struttura di minimizzazione alternata per il training di reti autoencoder e encoder-decoder", "Gli autori esplorano un approccio di ottimizzazione alternata per il training degli Auto Encoder, trattando ogni layer come un modello lineare generalizzato, e suggeriscono di usare la GD stocastica normalizzata come algoritmo di minimizzazione in ogni fase."]} +{"source": "We develop new algorithms for estimating heterogeneous treatment effects, combining recent developments in transfer learning for neural networks with insights from the causal inference literature. By taking advantage of transfer learning, we are able to efficiently use different data sources that are related to the same underlying causal mechanisms. We compare our algorithms with those in the extant literature using extensive simulation studies based on large-scale voter persuasion experiments and the MNIST database. Our methods can perform an order of magnitude better than existing benchmarks while using a fraction of the data.", "target": ["Transfer learning per stimare gli effetti causali usando le reti neurali.", "Sviluppa algoritmi per stimare l'effetto di trattamento medio condizionato per dataset ausiliari in diversi ambienti, sia con che senza learner di base.", "Gli autori propongono metodi per affrontare un nuovo task di transfer learning per stimare la funzione CATE, e li valutano utilizzando un setting sintetico e un dataset sperimentale del mondo reale.", "Utilizzando la regressione tramite rete neurale e confrontando i framework di transfer learning si stima un effetto di trattamento medio condizionato sotto l'ipotesi di ignorabilità delle stringhe"]} +{"source": "Neuronal assemblies, loosely defined as subsets of neurons with reoccurring spatio-temporally coordinated activation patterns, or \"motifs\", are thought to be building blocks of neural representations and information processing. We here propose LeMoNADe, a new exploratory data analysis method that facilitates hunting for motifs in calcium imaging videos, the dominant microscopic functional imaging modality in neurophysiology. Our nonparametric method extracts motifs directly from videos, bypassing the difficult intermediate step of spike extraction. Our technique augments variational autoencoders with a discrete stochastic node, and we show in detail how a differentiable reparametrization and relaxation can be used. An evaluation on simulated data, with available ground truth, reveals excellent quantitative performance. In real video data acquired from brain slices, with no ground truth available, LeMoNADe uncovers nontrivial candidate motifs that can help generate hypotheses for more focused biological investigations.", "target": ["Presentiamo LeMoNADe, un metodo di rilevamento di motif appresi end-to-end che opera direttamente sui video di imaging del calcio.", "Questo articolo propone un modello in stile VAE per identificare i motif dai video di imaging del calcio, basandosi sulle variabili di Bernouli e usando il trucco del Gumbel-softmax per l'inferenza."]} +{"source": "A noisy and diverse demonstration set may hinder the performances of an agent aiming to acquire certain skills via imitation learning. However, state-of-the-art imitation learning algorithms often assume the optimality of the given demonstration set. In this paper, we address such optimal assumption by learning only from the most suitable demonstrations in a given set. Suitability of a demonstration is estimated by whether imitating it produce desirable outcomes for achieving the goals of the tasks. For more efficient demonstration suitability assessments, the learning agent should be capable of imitating a demonstration as quick as possible, which shares similar spirit with fast adaptation in the meta-learning regime. Our framework, thus built on top of Model-Agnostic Meta-Learning, evaluates how desirable the imitated outcomes are, after adaptation to each demonstration in the set. The resulting assessments hence enable us to select suitable demonstration subsets for acquiring better imitated skills. The videos related to our experiments are available at: https://sites.google.com/view/deepdj", "target": ["Proponiamo una framework per imparare una buona policy attraverso l'imitation learning da un set di dimostrazioni rumorose attraverso il meta-training di un valutatore di idoneità della dimostrazione.", "Arricchisce l'imitation learning con un algoritmo basato su MAML che determina automaticamente se le dimostrazioni fornite sono \"adatte\".", "Un metodo per fare imitation learning da un insieme di dimostrazioni che include comportamenti inutili, che seleziona le dimostrazioni utili in base ai loro guadagni di performance forniti al momento del meta-training."]} +{"source": "We introduce causal implicit generative models (CiGMs): models that allow sampling from not only the true observational but also the true interventional distributions. We show that adversarial training can be used to learn a CiGM, if the generator architecture is structured based on a given causal graph. We consider the application of conditional and interventional sampling of face images with binary feature labels, such as mustache, young. We preserve the dependency structure between the labels with a given causal graph. We devise a two-stage procedure for learning a CiGM over the labels and the image. First we train a CiGM over the binary labels using a Wasserstein GAN where the generator neural network is consistent with the causal graph between the labels. Later, we combine this with a conditional GAN to generate images conditioned on the binary labels. We propose two new conditional GAN architectures: CausalGAN and CausalBEGAN. We show that the optimal generator of the CausalGAN, given the labels, samples from the image distributions conditioned on these labels. The conditional GAN combined with a trained CiGM for the labels is then a CiGM over the labels and the generated image. We show that the proposed architectures can be used to sample from observational and interventional image distributions, even for interventions which do not naturally occur in the dataset.", "target": ["Introduciamo modelli generativi causali impliciti, che possono campionare da distribuzioni condizionali e interventistiche e proponiamo anche due nuove GAN condizionali che usiamo per addestrarli.", "Un metodo per combinare un grafo casuale, che descrive la struttura di dipendenza delle label, con due architetture GAN condizionali che generano immagini condizionando sulla label binaria", "Gli autori affrontano la questione dell'apprendimento di un modello causale tra le variabili dell'immagine e l'immagine stessa dai dati osservazionali, quando è data una struttura causale tra le label dell'immagini."]} +{"source": "Self-normalizing discriminative models approximate the normalized probability of a class without having to compute the partition function. This property is useful to computationally-intensive neural network classifiers, as the cost of computing the partition function grows linearly with the number of classes and may become prohibitive. In particular, since neural language models may deal with up to millions of classes, their self-normalization properties received notable attention. Several recent studies empirically found that language models, trained using Noise Contrastive Estimation (NCE), exhibit self-normalization, but could not explain why. In this study, we provide a theoretical justification to this property by viewing NCE as a low-rank matrix approximation. Our empirical investigation compares NCE to the alternative explicit approach for self-normalizing language models. It also uncovers a surprising negative correlation between self-normalization and perplexity, as well as some regularity in the observed errors that may potentially be used for improving self-normalization algorithms in the future.", "target": ["Dimostriamo che NCE è auto-normalizzato e lo dimostriamo su dataset", "Presenta una prova dell'auto normalizzazione di NCE come risultato dell'essere un'approssimazione della matrice di basso rango della matrice delle probabilità condizionali normalizzate.", "Questo articolo considera il problema dei modelli auto-normalizzanti e spiega il meccanismo di auto-normalizzazione interpretando NCE in termini di fattorizzazione della matrice."]} +{"source": "Learning word representations from large available corpora relies on the distributional hypothesis that words present in similar contexts tend to have similar meanings. Recent work has shown that word representations learnt in this manner lack sentiment information which, fortunately, can be leveraged using external knowledge. Our work addresses the question: can affect lexica improve the word representations learnt from a corpus? In this work, we propose techniques to incorporate affect lexica, which capture fine-grained information about a word's psycholinguistic and emotional orientation, into the training process of Word2Vec SkipGram, Word2Vec CBOW and GloVe methods using a joint learning approach. We use affect scores from Warriner's affect lexicon to regularize the vector representations learnt from an unlabelled corpus. Our proposed method outperforms previously proposed methods on standard tasks for word similarity detection, outlier detection and sentiment detection. We also demonstrate the usefulness of our approach for a new task related to the prediction of formality, frustration and politeness in corporate communication.", "target": ["Arricchire gli embedding di parole con informazioni sugli affect migliora le loro prestazioni nei task di sentiment prediction.", "Propone di utilizzare i lexica degli affect per migliorare gli embedding di parole per superare lo standard Word2vec e Glove.", "Questo articolo propone di integrare le informazioni di una risorsa semantica che quantifica l'affect delle parole in un algoritmo di word embedding basato sul testo per rendere i modelli linguistici più riflessivi dei fenomeni semantici e pragmatici.", "Questo articolo introduce delle modifiche alle funzioni di loss word2vec e GloVe per incorporare i lexica degli affect e facilitare l'apprendimento di embedding di parole sensibili agli affect."]} +{"source": "Different kinds of representation learning techniques on graph have shown significant effect in downstream machine learning tasks. Recently, in order to inductively learn representations for graph structures that is unobservable during training, a general framework with sampling and aggregating (GraphSAGE) was proposed by Hamilton and Ying and had been proved more efficient than transductive methods on fileds like transfer learning or evolving dataset. However, GraphSAGE is uncapable of selective neighbor sampling and lack of memory of known nodes that've been trained. To address these problems, we present an unsupervised method that samples neighborhood information attended by co-occurring structures and optimizes a trainable global bias as a representation expectation for each node in the given graph. Experiments show that our approach outperforms the state-of-the-art inductive and unsupervised methods for representation learning on graphs.", "target": ["Per gli embedding non supervisionati e induttivi di reti, proponiamo un nuovo approccio per esplorare i vicini più rilevanti e preservare la conoscenza precedentemente appresa dei nodi utilizzando l'architettura di bi-attention e introducendo un bias globale, rispettivamente", "Questo paper propone un'estensione di GraphSAGE utilizzando una matrice globale di embedding di bias nelle funzioni di aggregazione locale e un metodo per campionare i nodi interessanti."]} +{"source": "Learning distributed representations for nodes in graphs is a crucial primitive in network analysis with a wide spectrum of applications. Linear graph embedding methods learn such representations by optimizing the likelihood of both positive and negative edges while constraining the dimension of the embedding vectors. We argue that the generalization performance of these methods is not due to the dimensionality constraint as commonly believed, but rather the small norm of embedding vectors. Both theoretical and empirical evidence are provided to support this argument: (a) we prove that the generalization error of these methods can be bounded by limiting the norm of vectors, regardless of the embedding dimension; (b) we show that the generalization performance of linear graph embedding methods is correlated with the norm of embedding vectors, which is small due to the early stopping of SGD and the vanishing gradients. We performed extensive experiments to validate our analysis and showcased the importance of proper norm regularization in practice.", "target": ["Noi sosteniamo che la generalizzazione dell'embedding dei grafi lineari non è dovuta al vincolo di dimensionalità ma piuttosto alla piccola norma dei vettori di embedding.", "Gli autori mostrano che l'errore di generalizzazione dei metodi di embedding dei grafi lineari è limitato dalla norma dei vettori di embedding piuttosto che dai vincoli di dimensionalità", "Gli autori propongono un limite teorico sulle prestazioni di generalizzazione dell'apprendimento di embedding di grafi e sostengono che la norma delle coordinate determina il successo della rappresentazione appresa."]} +{"source": "Momentum-based acceleration of stochastic gradient descent (SGD) is widely used in deep learning. We propose the quasi-hyperbolic momentum algorithm (QHM) as an extremely simple alteration of momentum SGD, averaging a plain SGD step with a momentum step. We describe numerous connections to and identities with other algorithms, and we characterize the set of two-state optimization algorithms that QHM can recover. Finally, we propose a QH variant of Adam called QHAdam, and we empirically demonstrate that our algorithms lead to significantly improved training in a variety of settings, including a new state-of-the-art result on WMT16 EN-DE. We hope that these empirical results, combined with the conceptual and practical simplicity of QHM and QHAdam, will spur interest from both practitioners and researchers. Code is immediately available.", "target": ["Mescolate SGD e momentum (o fate qualcosa di simile con Adam) per ottenere un grande profitto.", "L'articolo propone semplici modifiche a SGD e Adam, chiamate varianti QH, che possono recuperare il metodo originale e una serie di altri trucchi di ottimizzazione.", "Una variante del momentum classico che prende una media ponderata del momentum e dell'aggiornamento del gradiente, e una valutazione delle sue relazioni con altri schemi di ottimizzazione basati sul momentum."]} +{"source": "Reinforcement Learning (RL) can model complex behavior policies for goal-directed sequential decision making tasks. A hallmark of RL algorithms is Temporal Difference (TD) learning: value function for the current state is moved towards a bootstrapped target that is estimated using the next state's value function. lambda-returns define the target of the RL agent as a weighted combination of rewards estimated by using multiple many-step look-aheads. Although mathematically tractable, the use of exponentially decaying weighting of n-step returns based targets in lambda-returns is a rather ad-hoc design choice. Our major contribution is that we propose a generalization of lambda-returns called Confidence-based Autodidactic Returns (CAR), wherein the RL agent learns the weighting of the n-step returns in an end-to-end manner. In contrast to lambda-returns wherein the RL agent is restricted to use an exponentially decaying weighting scheme, CAR allows the agent to learn to decide how much it wants to weigh the n-step returns based targets. Our experiments, in addition to showing the efficacy of CAR, also empirically demonstrate that using sophisticated weighted mixtures of multi-step returns (like CAR and lambda-returns) considerably outperforms the use of n-step returns. We perform our experiments on the Asynchronous Advantage Actor Critic (A3C) algorithm in the Atari 2600 domain.", "target": ["Un nuovo modo di generalizzare i lambda-return permettendo all'agente RL di decidere quanto vuole pesare ciascuno degli n-step return.", "Estende l'algoritmo A3C con lambda return, e propone un approccio per imparare i pesi dei return.", "Gli autori presentano i return autodidattici confidence-based, un metodo di deep RL per regolare i pesi di un vettore di ammissibilità nella stima del valore TD(lambda)-like per favorire stime più stabili dello stato."]} +{"source": "Current end-to-end deep learning driving models have two problems: (1) Poor generalization ability of unobserved driving environment when diversity of train- ing driving dataset is limited (2) Lack of accident explanation ability when driving models don’t work as expected. To tackle these two problems, rooted on the be- lieve that knowledge of associated easy task is benificial for addressing difficult task, we proposed a new driving model which is composed of perception module for see and think and driving module for behave, and trained it with multi-task perception-related basic knowledge and driving knowledge stepwisely. Specifi- cally segmentation map and depth map (pixel level understanding of images) were considered as what & where and how far knowledge for tackling easier driving- related perception problems before generating final control commands for difficult driving task. The results of experiments demonstrated the effectiveness of multi- task perception knowledge for better generalization and accident explanation abil- ity. With our method the average sucess rate of finishing most difficult navigation tasks in untrained city of CoRL test surpassed current benchmark method for 15 percent in trained weather and 20 percent in untrained weathers.", "target": ["Abbiamo proposto un nuovo modello di guida autonoma che è composto da modulo di percezione per vedere e pensare e un modulo di guida per comportarsi al fine di acquisire una migliore generalizzazione e capacità di spiegazione degli incidenti.", "Presenta un'architettura di multitask learning per la stima della mappa di profondità, segmentazione e la predizione di guida utilizzando un modulo di percezione e un modulo di decisione di guida.", "Un metodo per un'architettura end-to-end modificata che ha una migliore generalizzazione e capacità di spiegazione, è più robusto per un diverso setting di test e ha un output di decodifica che può aiutare nel debugging del modello.", "Gli autori presentano una rete neurale convoluzionale multi-task per la guida end-to-end e forniscono valutazioni con il simulatore open source CARLA mostrando migliori prestazioni di generalizzazione in nuove condizioni di guida rispetto alle baseline"]} +{"source": "Recently there has been a surge of interest in designing graph embedding methods. Few, if any, can scale to a large-sized graph with millions of nodes due to both computational complexity and memory requirements. In this paper, we relax this limitation by introducing the MultI-Level Embedding (MILE) framework – a generic methodology allowing contemporary graph embedding methods to scale to large graphs. MILE repeatedly coarsens the graph into smaller ones using a hybrid matching technique to maintain the backbone structure of the graph. It then applies existing embedding methods on the coarsest graph and refines the embeddings to the original graph through a novel graph convolution neural network that it learns. The proposed MILE framework is agnostic to the underlying graph embedding techniques and can be applied to many existing graph embedding methods without modifying them. We employ our framework on several popular graph embedding techniques and conduct embedding for real-world graphs. Experimental results on five large-scale datasets demonstrate that MILE significantly boosts the speed (order of magnitude) of graph embedding while also often generating embeddings of better quality for the task of node classification. MILE can comfortably scale to a graph with 9 million nodes and 40 million edges, on which existing methods run out of memory or take too long to compute on a modern workstation.", "target": ["Un framework generico per scalare le tecniche esistenti di embedding di grafi a grandi grafi.", "Questo articolo propone un framework di embedding multilivello da applicare sopra i metodi di embedding di rete esistenti al fine di scalare a reti su larga scala con maggiore velocità.", "Gli autori propongono un framework a tre stadi per l'embedding di grafi su larga scala con una migliore qualità di embedding."]} +{"source": "Anomaly detection discovers regular patterns in unlabeled data and identifies the non-conforming data points, which in some cases are the result of malicious attacks by adversaries. Learners such as One-Class Support Vector Machines (OCSVMs) have been successfully in anomaly detection, yet their performance may degrade significantly in the presence of sophisticated adversaries, who target the algorithm itself by compromising the integrity of the training data. With the rise in the use of machine learning in mission critical day-to-day activities where errors may have significant consequences, it is imperative that machine learning systems are made secure. To address this, we propose a defense mechanism that is based on a contraction of the data, and we test its effectiveness using OCSVMs. The proposed approach introduces a layer of uncertainty on top of the OCSVM learner, making it infeasible for the adversary to guess the specific configuration of the learner. We theoretically analyze the effects of adversarial perturbations on the separating margin of OCSVMs and provide empirical evidence on several benchmark datasets, which show that by carefully contracting the data in low dimensional spaces, we can successfully identify adversarial samples that would not have been identifiable in the original dimensional space. The numerical results show that the proposed method improves OCSVMs performance significantly (2-7%)", "target": ["Un nuovo metodo per aumentare la resistenza degli OCSVM contro attacchi mirati e di integrità mediante trasformazioni selettive non lineari dei dati a dimensioni inferiori.", "Gli autori propongono una difesa contro gli attacchi alla sicurezza dei rilevatori di anomalie basati su un SVM one-class", "Questo articolo esplora come le proiezioni casuali possono essere utilizzate per rendere OCSVM robusta ai dati di allenamento con adversarial perturbation."]} +{"source": "In this paper, we present a layer-wise learning of stochastic neural networks (SNNs) in an information-theoretic perspective. In each layer of an SNN, the compression and the relevance are defined to quantify the amount of information that the layer contains about the input space and the target space, respectively. We jointly optimize the compression and the relevance of all parameters in an SNN to better exploit the neural network's representation. Previously, the Information Bottleneck (IB) framework (\\cite{Tishby99}) extracts relevant information for a target variable. Here, we propose Parametric Information Bottleneck (PIB) for a neural network by utilizing (only) its model parameters explicitly to approximate the compression and the relevance. We show that, as compared to the maximum likelihood estimate (MLE) principle, PIBs : (i) improve the generalization of neural networks in classification tasks, (ii) push the representation of neural networks closer to the optimal information-theoretical representation in a faster manner.", "target": ["Apprendimento di una migliore rappresentazione delle reti neurali con il principio dell'Information Bottleneck", "Propone un metodo di apprendimento basato sul framework dell'information bottleneck, dove gli hidden layer delle deep network comprimono l'input X mantenendo informazioni sufficienti per prevedere l'output Y.", "Questo articolo presenta un nuovo modo di addestrare una rete neurale stocastica seguendo un framework di rilevanza/compressione delle informazioni simile all'Information Bottleneck."]} +{"source": "The maximum mean discrepancy (MMD) between two probability measures P and Q is a metric that is zero if and only if all moments of the two measures are equal, making it an appealing statistic for two-sample tests. Given i.i.d. samples from P and Q, Gretton et al. (2012) show that we can construct an unbiased estimator for the square of the MMD between the two distributions. If P is a distribution of interest and Q is the distribution implied by a generative neural network with stochastic inputs, we can use this estimator to train our neural network. However, in practice we do not always have i.i.d. samples from our target of interest. Data sets often exhibit biases—for example, under-representation of certain demographics—and if we ignore this fact our machine learning algorithms will propagate these biases. Alternatively, it may be useful to assume our data has been gathered via a biased sample selection mechanism in order to manipulate properties of the estimating distribution Q. In this paper, we construct an estimator for the MMD between P and Q when we only have access to P via some biased sample selection mechanism, and suggest methods for estimating this sample selection mechanism when it is not already known. We show that this estimator can be used to train generative neural networks on a biased data sample, to give a simulator that reverses the effect of that bias.", "target": ["Proponiamo uno stimatore per la massima discrepanza media, appropriato quando una distribuzione di destinazione è accessibile solo attraverso una procedura di sample selection biased, e mostriamo che può essere utilizzato in una rete generativa per correggere questo bias.", "Propone uno stimatore importance-weighted della MMD per stimare la MMD tra distribuzioni basate su sample biased secondo uno schema noto o stimato.", "Gli autori affrontano il problema del bias della selezione del sample nelle MMD-GAN e propongono una stima della MMD tra due distribuzioni utilizzando la discrepanza media massima pesata.", "Questo articolo presenta una modifica dell'obiettivo utilizzato per addestrare le reti generative con un avversario MMD"]} +{"source": "We propose Bayesian Deep Q-Network (BDQN), a practical Thompson sampling based Reinforcement Learning (RL) Algorithm. Thompson sampling allows for targeted exploration in high dimensions through posterior sampling but is usually computationally expensive. We address this limitation by introducing uncertainty only at the output layer of the network through a Bayesian Linear Regression (BLR) model, which can be trained with fast closed-form updates and its samples can be drawn efficiently through the Gaussian distribution. We apply our method to a wide range of Atari Arcade Learning Environments. Since BDQN carries out more efficient exploration, it is able to reach higher rewards substantially faster than a key baseline, DDQN.", "target": ["Si usa la regressione bayesiana per stimare il posteriore sulle funzioni Q e si utilizza il sampling di Thompson come una strategia di esplorazione mirata con un efficiente trade-off tra exploration e exploitation", "Gli autori propongono un nuovo algoritmo per l'esplorazione in Deep RL dove applicano la regressione lineare bayesiana con le feature dell'ultimo layer di una rete DQN per stimare la funzione Q per ogni azione.", "Gli autori descrivono come usare le reti neurali bayesiane con il sampling di Thompson per un'esplorazione efficiente in q-learning e propongono un approccio che supera gli approcci di esplorazione epsilon-greedy."]} +{"source": "In this work, we propose the polynomial convolutional neural network (PolyCNN), as a new design of a weight-learning efficient variant of the traditional CNN. The biggest advantage of the PolyCNN is that at each convolutional layer, only one convolutional filter is needed for learning the weights, which we call the seed filter, and all the other convolutional filters are the polynomial transformations of the seed filter, which is termed as an early fan-out. Alternatively, we can also perform late fan-out on the seed filter response to create the number of response maps needed to be input into the next layer. Both early and late fan-out allow the PolyCNN to learn only one convolutional filter at each layer, which can dramatically reduce the model complexity by saving 10x to 50x parameters during learning. While being efficient during both training and testing, the PolyCNN does not suffer performance due to the non-linear polynomial expansion which translates to richer representational power within the convolutional layers. By allowing direct control over model complexity, PolyCNN provides a flexible trade-off between performance and efficiency. We have verified the on-par performance between the proposed PolyCNN and the standard CNN on several visual datasets, such as MNIST, CIFAR-10, SVHN, and ImageNet.", "target": ["PolyCNN ha solo bisogno di imparare un filtro convoluzionario seed ad ogni layer. Questa è una variante efficiente della CNN tradizionale, con prestazioni alla pari.", "I tentativi di ridurre il numero di parametri del modello CNN utilizzando la trasformazione polinomiale dei filtri per creare blow-up delle risposte dei filtri.", "Gli autori propongono un'architettura di weight sharing per ridurre il numero di parametri delle reti neurali convoluzionali con seed filter"]} +{"source": "Detecting the emergence of abrupt property changes in time series is a challenging problem. Kernel two-sample test has been studied for this task which makes fewer assumptions on the distributions than traditional parametric approaches. However, selecting kernels is non-trivial in practice. Although kernel selection for the two-sample test has been studied, the insufficient samples in change point detection problem hinder the success of those developed kernel selection algorithms. In this paper, we propose KL-CPD, a novel kernel learning framework for time series CPD that optimizes a lower bound of test power via an auxiliary generative model. With deep kernel parameterization, KL-CPD endows kernel two-sample test with the data-driven kernel to detect different types of change-points in real-world applications. The proposed approach significantly outperformed other state-of-the-art methods in our comparative evaluation of benchmark datasets and simulation studies.", "target": ["In questo articolo, proponiamo KL-CPD, un nuovo framework di kernel learning per la CPD di serie temporali che ottimizza un limite inferiore del test power attraverso un modello generativo ausiliario come surrogato della distribuzione anomala.", "Descrive un nuovo approccio per ottimizzare la scelta del kernel verso una maggiore test power e dimostra di offrire miglioramenti rispetto alle alternative."]} +{"source": "Theories in cognitive psychology postulate that humans use similarity as a basis for object categorization. However, work in image classification generally as- sumes disjoint and equally dissimilar classes to achieve super-human levels of performance on certain datasets. In our work, we adapt notions of similarity using weak labels over multiple hierarchical levels to boost classification performance. Instead of pitting clustering directly against classification, we use a warm-start based evaluation to explicitly provide value to a clustering representation by its ability to aid classification. We evaluate on CIFAR10 and a fine-grained classifi- cation dataset to show improvements in performance with the procedural addition of intermediate losses and weak labels based on multiple hierarchy levels. Further- more, we show that pretraining AlexNet on hierarchical weak labels in conjunc- tion with intermediate losses outperforms a classification baseline by over 17% on a subset of Birdsnap dataset. Finally, we show improvement over AlexNet trained using ImageNet pre-trained weights as initializations which further supports our claim of the importance of similarity.", "target": ["Cluster prima di classificare; usare weak label per migliorare la classificazione ", "Propone l'uso di una loss basata sul clustering a più livelli di una deepnet e l'uso della struttura gerarchica dello spazio delle label per allenare rappresentazioni migliori.", "Questo articolo usa le informazioni gerarchiche sulle label per imporre ulteriori loss sulle rappresentazioni intermedie nel training delle reti neurali."]} +{"source": "Deep reinforcement learning algorithms that estimate state and state-action value functions have been shown to be effective in a variety of challenging domains, including learning control strategies from raw image pixels. However, algorithms that estimate state and state-action value functions typically assume a fully observed state and must compensate for partial or non-Markovian observations by using finite-length frame-history observations or recurrent networks. In this work, we propose a new deep reinforcement learning algorithm based on counterfactual regret minimization that iteratively updates an approximation to a cumulative clipped advantage function and is robust to partially observed state. We demonstrate that on several partially observed reinforcement learning tasks, this new class of algorithms can substantially outperform strong baseline methods: on Pong with single-frame observations, and on the challenging Doom (ViZDoom) and Minecraft (Malmö) first-person navigation benchmarks.", "target": ["Advantage-based regret minimization è un nuovo algoritmo di deep reinforcement learning che è particolarmente efficace su task parzialmente osservabili, come la navigazione in prima persona in Doom e Minecraft.", "Questo articolo introduce i concetti di minimizzazione del counterfactual regret nel campo del deep RL e un algoritmo chiamato ARM che può trattare meglio l'osservabilità parziale.", "L'articolo fornisce una variante ispirata alla teoria dei giochi dell'algoritmo policy-gradient basata sull'idea di minimizzazione del counterfactual regret e sostiene che l'approccio può trattare il dominio parziale osservabile meglio dei metodi standard."]} +{"source": "Recent deep multi-task learning (MTL) has been witnessed its success in alleviating data scarcity of some task by utilizing domain-specific knowledge from related tasks. Nonetheless, several major issues of deep MTL, including the effectiveness of sharing mechanisms, the efficiency of model complexity and the flexibility of network architectures, still remain largely unaddressed. To this end, we propose a novel generalized latent-subspace based knowledge sharing mechanism for linking task-specific models, namely tensor ring multi-task learning (TRMTL). TRMTL has a highly compact representation, and it is very effective in transferring task-invariant knowledge while being super flexible in learning task-specific features, successfully mitigating the dilemma of both negative-transfer in lower layers and under-transfer in higher layers. Under our TRMTL, it is feasible for each task to have heterogenous input data dimensionality or distinct feature sizes at different hidden layers. Experiments on a variety of datasets demonstrate our model is capable of significantly improving each single task’s performance, particularly favourable in scenarios where some of the tasks have insufficient data.", "target": ["Un modello di deep multi-task learning che adatta la rappresentazione dell'anello dei tensori", "Una variante della formulazione dell'anello dei tensori per multi-task learning condividendo alcuni dei nuclei TT per l'apprendimento del \"task comune\" mentre si imparano singoli nuclei TT per ogni task separato"]} +{"source": "Neural Processes (NPs) (Garnelo et al., 2018) approach regression by learning to map a context set of observed input-output pairs to a distribution over regression functions. Each function models the distribution of the output given an input, conditioned on the context. NPs have the benefit of fitting observed data efficiently with linear complexity in the number of context input-output pairs, and can learn a wide family of conditional distributions; they learn predictive distributions conditioned on context sets of arbitrary size. Nonetheless, we show that NPs suffer a fundamental drawback of underfitting, giving inaccurate predictions at the inputs of the observed data they condition on. We address this issue by incorporating attention into NPs, allowing each input location to attend to the relevant context points for the prediction. We show that this greatly improves the accuracy of predictions, results in noticeably faster training, and expands the range of functions that can be modelled.", "target": ["Un modello di regressione che impara le distribuzioni condizionali di un processo stocastico, incorporando l'attention nei processi neurali.", "Propone di risolvere il problema dell'underfitting nel metodo del processo neurale aggiungendo un meccanismo di attention al percorso deterministico.", "Un'estensione del framework dei Processi Neurali che aggiunge un meccanismo di condizionamento basato sull'attention, permettendo al modello di catturare meglio le dipendenze nel conditioning set.", "Gli autori estendono i processi neurali incorporando la self-attention per arricchire le feature dei punti di contesto e la cross-attention per produrre una rappresentazione specifica della query. Risolvono il problema dell'underfitting delle NP e mostrano che le ANP convergono meglio e più velocemente delle NP."]} +{"source": "Deconvolutional layers have been widely used in a variety of deep models for up-sampling, including encoder-decoder networks for semantic segmentation and deep generative models for unsupervised learning. One of the key limitations of deconvolutional operations is that they result in the so-called checkerboard problem. This is caused by the fact that no direct relationship exists among adjacent pixels on the output feature map. To address this problem, we propose the pixel deconvolutional layer (PixelDCL) to establish direct relationships among adjacent pixels on the up-sampled feature map. Our method is based on a fresh interpretation of the regular deconvolution operation. The resulting PixelDCL can be used to replace any deconvolutional layer in a plug-and-play manner without compromising the fully trainable capabilities of original models. The proposed PixelDCL may result in slight decrease in efficiency, but this can be overcome by an implementation trick. Experimental results on semantic segmentation demonstrate that PixelDCL can consider spatial features such as edges and shapes and yields more accurate segmentation outputs than deconvolutional layers. When used in image generation tasks, our PixelDCL can largely overcome the checkerboard problem suffered by regular deconvolution operations.", "target": ["Risolvere il problema della scacchiera nel layer di deconvoluzione costruendo dipendenze tra i pixel", "Questo lavoro propone layer di deconvoluzione dei pixel per le reti neurali convoluzionali come un modo per alleviare l'effetto scacchiera.", "Una nuova tecnica per generalizzare le operazioni di deconvoluzione utilizzate nelle architetture CNN standard, che propone di fare la predizione sequenziale delle feature dei pixel adiacenti, ottenendo output più spazialmente smooth per i layer di deconvoluzione."]} +{"source": "In this paper, the preparation of a neural network for pruning and few-bit quantization is formulated as a variational inference problem. To this end, a quantizing prior that leads to a multi-modal, sparse posterior distribution over weights, is introduced and a differentiable Kullback-Leibler divergence approximation for this prior is derived. After training with Variational Network Quantization, weights can be replaced by deterministic quantization values with small to negligible loss of task accuracy (including pruning by setting weights to 0). The method does not require fine-tuning after quantization. Results are shown for ternary quantization on LeNet-5 (MNIST) and DenseNet (CIFAR-10).", "target": ["Quantizziamo e facciamo pruning dei pesi delle reti neurali usando l'inferenza bayesiana variazionale con un prior multimodale che induce la sparsità.", "Propone di utilizzare una miscela di spike continui come prior per una rete neurale bayesiana e dimostra le buone prestazioni con convnet relativamente sparsificate per minst e cifar-10.", "Questo articolo presenta un approccio variazionale bayesiano per quantizzare i pesi delle reti neurali a valori ternari dopo il training."]} +{"source": "Deep neural networks (DNNs) although achieving human-level performance in many domains, have very large model size that hinders their broader applications on edge computing devices. Extensive research work have been conducted on DNN model compression or pruning. However, most of the previous work took heuristic approaches. This work proposes a progressive weight pruning approach based on ADMM (Alternating Direction Method of Multipliers), a powerful technique to deal with non-convex optimization problems with potentially combinatorial constraints. Motivated by dynamic programming, the proposed method reaches extremely high pruning rate by using partial prunings with moderate pruning rates. Therefore, it resolves the accuracy degradation and long convergence time problems when pursuing extremely high pruning ratios. It achieves up to 34× pruning rate for ImageNet dataset and 167× pruning rate for MNIST dataset, significantly higher than those reached by the literature work. Under the same number of epochs, the proposed method also achieves faster convergence and higher compression rates. The codes and pruned DNN models are released in the anonymous link bit.ly/2zxdlss.", "target": ["Implementiamo un approccio di pruning dei pesi di DNN che raggiunge i più alti tassi di pruning.", "Questo articolo si concentra sul pruning dei pesi per la compressione delle reti neurali, raggiungendo un tasso di compressione pari a 30 per AlexNet e VGG su ImageNet.", "Una tecnica di pruning progressivo che impone un vincolo di sparsità strutturale sul parametro del peso e riscrive l'ottimizzazione come un framework ADMM, ottenendo una maggiore accuratezza rispetto alla projected gradient descent."]} +{"source": "In this paper, we present a new deep learning architecture for addressing the problem of supervised learning with sparse and irregularly sampled multivariate time series. The architecture is based on the use of a semi-parametric interpolation network followed by the application of a prediction network. The interpolation network allows for information to be shared across multiple dimensions of a multivariate time series during the interpolation stage, while any standard deep learning model can be used for the prediction network. This work is motivated by the analysis of physiological time series data in electronic health records, which are sparse, irregularly sampled, and multivariate. We investigate the performance of this architecture on both classification and regression tasks, showing that our approach outperforms a range of baseline and recently proposed models.", "target": ["Questo articolo presenta una nuova architettura di deep learning per affrontare il problema dell'apprendimento supervisionato con serie temporali multivariate sparse e campionate in modo irregolare.", "Propone una framework per fare predizioni su dati di serie temporali sparsi e campionati in modo irregolare usando un modulo di interpolazione che modella i valori mancanti usando interpolazione smooth, interpolazione non smooth e intensità.", "Risolve il problema dell'apprendimento supervisionato con serie temporali multivariate sparse e campionate in modo irregolare usando una rete di interpolazione semi-parametrica seguita da una rete di predizione."]} +{"source": "We introduce an analytic distance function for moderately sized point sets of known cardinality that is shown to have very desirable properties, both as a loss function as well as a regularizer for machine learning applications. We compare our novel construction to other point set distance functions and show proof of concept experiments for training neural networks end-to-end on point set prediction tasks such as object detection.", "target": ["Loss invariante alla permutazione per la predizione di set di punti.", "Propone una nuova loss per la point registration (allineamento di due insiemi di punti) con una proprietà desiderabile di invarianza alla permutazione.", "Questo articolo introduce una nuova funzione di distanza tra insiemi di punti, applica altre due distanze di permutazione in un task di object detection end-to-end, e mostra che in due dimensioni tutti i minimi locali della loss olografica sono minimi globali.", "Propone funzioni di loss invarianti alla permutazione che dipendono dalla distanza degli insiemi."]} +{"source": "We introduce a hierarchical model for efficient placement of computational graphs onto hardware devices, especially in heterogeneous environments with a mixture of CPUs, GPUs, and other computational devices. Our method learns to assign graph operations to groups and to allocate those groups to available devices. The grouping and device allocations are learned jointly. The proposed method is trained with policy gradient and requires no human intervention. Experiments with widely-used computer vision and natural language models show that our algorithm can find optimized, non-trivial placements for TensorFlow computational graphs with over 80,000 operations. In addition, our approach outperforms placements by human experts as well as a previous state-of-the-art placement method based on deep reinforcement learning. Our method achieves runtime reductions of up to 60.6% per training step when applied to models such as Neural Machine Translation.", "target": ["Introduciamo un modello gerarchico per il posizionamento efficiente, end-to-end, di grafi computazionali su dispositivi hardware.", "Propone di apprendere congiuntamente gruppi di operatori per collocolare e piazzare i gruppi appresi sui dispositivi per distribuire le operazioni per il deep learning tramite reinforcement learning.", "Gli autori hanno lo scopo di presentare una rete fully connected per sostituire lo step di co-location in un metodo di auto-placement proposto per accelerare il tempo di esecuzione di un modello su TensorFlow.", "Propone un algoritmo di device placement per collocare le operazioni di tensorflow sui dispositivi."]} +{"source": "Motion is an important signal for agents in dynamic environments, but learning to represent motion from unlabeled video is a difficult and underconstrained problem. We propose a model of motion based on elementary group properties of transformations and use it to train a representation of image motion. While most methods of estimating motion are based on pixel-level constraints, we use these group properties to constrain the abstract representation of motion itself. We demonstrate that a deep neural network trained using this method captures motion in both synthetic 2D sequences and real-world sequences of vehicle motion, without requiring any labels. Networks trained to respect these constraints implicitly identify the image characteristic of motion in different sequence types. In the context of vehicle motion, this method extracts information useful for localization, tracking, and odometry. Our results demonstrate that this representation is useful for learning motion in the general setting where explicit labels are difficult to obtain.", "target": ["Proponiamo un metodo per usare le proprietà del gruppo per imparare una rappresentazione del movimento senza label e dimostriamo l'uso di questo metodo per rappresentare il movimento 2D e 3D.", "Propone di imparare il gruppo di movimento rigido da una rappresentazione latente di sequenze di immagini senza la necessità di label esplicite e dimostra sperimentalmente il metodo su sequenze di cifre MINST e il dataset KITTI.", "Questo articolo propone un approccio per l'apprendimento delle feature del movimento video in modo unsupervised, utilizzando i vincoli per ottimizzare la rete neurale per produrre feature che possono essere utilizzate per fare regression per l'odometria."]} +{"source": "This paper introduces the concept of continuous convolution to neural networks and deep learning applications in general. Rather than directly using discretized information, input data is first projected into a high-dimensional Reproducing Kernel Hilbert Space (RKHS), where it can be modeled as a continuous function using a series of kernel bases. We then proceed to derive a closed-form solution to the continuous convolution operation between two arbitrary functions operating in different RKHS. Within this framework, convolutional filters also take the form of continuous functions, and the training procedure involves learning the RKHS to which each of these filters is projected, alongside their weight parameters. This results in much more expressive filters, that do not require spatial discretization and benefit from properties such as adaptive support and non-stationarity. Experiments on image classification are performed, using classical datasets, with results indicating that the proposed continuous convolutional neural network is able to achieve competitive accuracy rates with far fewer parameters and a faster convergence rate.", "target": ["Questo articolo propone un nuovo layer di convoluzione che opera in uno spazio di Hilbert continuo a kernel riproducente.", "Proiettare gli esempi in uno spazio RK Hilbert ed eseguire la convoluzione e il filtraggio in quello spazio.", "Questo articolo formula una variante delle reti neurali convoluzionali che modella sia le attivazioni che i filtri come funzioni continue composte da basi di kernel"]} +{"source": "Convolutional Neural Networks (CNNs) are commonly thought to recognise objects by learning increasingly complex representations of object shapes. Some recent studies suggest a more important role of image textures. We here put these conflicting hypotheses to a quantitative test by evaluating CNNs and human observers on images with a texture-shape cue conflict. We show that ImageNet-trained CNNs are strongly biased towards recognising textures rather than shapes, which is in stark contrast to human behavioural evidence and reveals fundamentally different classification strategies. We then demonstrate that the same standard architecture (ResNet-50) that learns a texture-based representation on ImageNet is able to learn a shape-based representation instead when trained on 'Stylized-ImageNet', a stylized version of ImageNet. This provides a much better fit for human behavioural performance in our well-controlled psychophysical lab setting (nine experiments totalling 48,560 psychophysical trials across 97 observers) and comes with a number of unexpected emergent benefits such as improved object detection performance and previously unseen robustness towards a wide range of image distortions, highlighting advantages of a shape-based representation.", "target": ["Le CNN addestrate in ImageNet sono orientate verso la texture dell'oggetto (invece che verso la forma come gli umani). Il superamento di questa grande differenza tra la visione umana e la visione della macchina produce prestazioni di rilevamento migliorate e una robustezza mai vista prima alle distorsioni dell'immagine.", "Usare lo stylizaton delle immagini per aumentare i dati di training per le CNN addestrate in ImageNet per far apparire le reti risultanti più allineate ai giudizi umani", "Questo articolo studia le CNN come AlexNet, VGG, GoogleNet e ResNet50, mostra che questi modelli sono biased verso le texture quando addestrati su ImageNet, e propone un nuovo dataset ImageNet."]} +{"source": "In this work, we exploited different strategies to provide prior knowledge to commonly used generative modeling approaches aiming to obtain speaker-dependent low dimensional representations from short-duration segments of speech data, making use of available information of speaker identities. Namely, convolutional variational autoencoders are employed, and statistics of its learned posterior distribution are used as low dimensional representations of fixed length short-duration utterances. In order to enforce speaker dependency in the latent layer, we introduced a variation of the commonly used prior within the variational autoencoders framework, i.e. the model is simultaneously trained for reconstruction of inputs along with a discriminative task performed on top of latent layers outputs. The effectiveness of both triplet loss minimization and speaker recognition are evaluated as implicit priors on the challenging cross-language NIST SRE 2016 setting and compared against fully supervised and unsupervised baselines.", "target": ["Valutiamo l'efficacia di avere task discriminatori ausiliari eseguiti sopra le statistiche della distribuzione posteriore appresa dagli autoencoder variazionali per imporre la speaker dependency.", "Proporre un modello di autoencoder per imparare una rappresentazione per la speaker verification utilizzando finestre di analisi di breve durata.", "Una versione modificata del modello di autoencoder variazionale che affronta il problema dello speaker recognition nel contesto di segmenti di breve durata"]} +{"source": "The importance-weighted autoencoder (IWAE) approach of Burda et al. defines a sequence of increasingly tighter bounds on the marginal likelihood of latent variable models. Recently, Cremer et al. reinterpreted the IWAE bounds as ordinary variational evidence lower bounds (ELBO) applied to increasingly accurate variational distributions. In this work, we provide yet another perspective on the IWAE bounds. We interpret each IWAE bound as a biased estimator of the true marginal likelihood where for the bound defined on $K$ samples we show the bias to be of order O(1/K). In our theoretical analysis of the IWAE objective we derive asymptotic bias and variance expressions. Based on this analysis we develop jackknife variational inference (JVI), a family of bias-reduced estimators reducing the bias to $O(K^{-(m+1)})$ for any given m < K while retaining computational efficiency. Finally, we demonstrate that JVI leads to improved evidence estimates in variational autoencoders. We also report first results on applying JVI to learning variational autoencoders. Our implementation is available at https://github.com/Microsoft/jackknife-variational-inference", "target": ["L'inferenza variazionale è biased, facciamo debiasing.", "Introduce l'inferenza variazionale jackknife, un metodo per il debiasing di obiettivi Monte Carlo come l'auto-encoder importance weighted.", "Gli autori analizzano il bias e la varianza del limite IWAE e derivano un approccio jacknife per stimare i momenti come un modo per fare debiasing dell'IWAE per sample pesati con importanza finita."]} +{"source": "In this paper, we consider the problem of autonomous lane changing for self driving vehicles in a multi-lane, multi-agent setting. We present a framework that demonstrates a more structured and data efficient alternative to end-to-end complete policy learning on problems where the high-level policy is hard to formulate using traditional optimization or rule based methods but well designed low-level controllers are available. Our framework uses deep reinforcement learning solely to obtain a high-level policy for tactical decision making, while still maintaining a tight integration with the low-level controller, thus getting the best of both worlds. We accomplish this with Q-masking, a technique with which we are able to incorporate prior knowledge, constraints, and information from a low-level controller, directly in to the learning process thereby simplifying the reward function and making learning faster and data efficient. We provide preliminary results in a simulator and show our approach to be more efficient than a greedy baseline, and more successful and safer than human driving.", "target": ["Un framework che fornisce una policy per il cambio di corsia autonomo imparando a prendere decisioni tattiche di alto livello con il deep reinforcement learning, e mantenendo una stretta integrazione con un controller di basso livello per prendere azioni di basso livello.", "Considera il problema del cambio di corsia autonomo per le auto a guida autonoma nel setting multi-lane multi-agente slot car, propone una nuova strategia di apprendimento Q-masking - accoppiando un controller definito di basso livello con una policy decisionale tattica di alto livello.", "Questo articolo propone un approccio di deep Q-learning al problema del cambio di corsia usando il \"Q-masking\", che riduce lo spazio d'azione in base a vincoli o conoscenze precedenti.", "Gli autori propongono un metodo che utilizza una policy di alto livello basata sul Q-learning che è combinata con una maschera contestuale derivata dai vincoli di sicurezza e dai controllori di basso livello, che disabilitano certe azioni dall'essere selezionabili in certi stati."]} +{"source": "Despite the recent successes in robotic locomotion control, the design of robot relies heavily on human engineering. Automatic robot design has been a long studied subject, but the recent progress has been slowed due to the large combinatorial search space and the difficulty in evaluating the found candidates. To address the two challenges, we formulate automatic robot design as a graph search problem and perform evolution search in graph space. We propose Neural Graph Evolution (NGE), which performs selection on current candidates and evolves new ones iteratively. Different from previous approaches, NGE uses graph neural networks to parameterize the control policies, which reduces evaluation cost on new candidates with the help of skill transfer from previously evaluated designs. In addition, NGE applies Graph Mutation with Uncertainty (GM-UC) by incorporating model uncertainty, which reduces the search space by balancing exploration and exploitation. We show that NGE significantly outperforms previous methods by an order of magnitude. As shown in experiments, NGE is the first algorithm that can automatically discover kinematically preferred robotic graph structures, such as a fish with two symmetrical flat side-fins and a tail, or a cheetah with athletic front and back legs. Instead of using thousands of cores for weeks, NGE efficiently solves searching problem within a day on a single 64 CPU-core Amazon EC2 machine.", "target": ["Ricerca automatica di design robotico con graph neural network", "Propone un approccio per la progettazione automatica di robot basato sull'evoluzione del grafo neurale. Gli esperimenti dimostrano che ottimizzare sia il controller che l'hardware è meglio che ottimizzare solo il controller.", "Gli autori propongono uno schema basato su una rappresentazione a grafo della struttura del robot, e una graph neural network come controllore per ottimizzare le strutture del robot, combinate con i loro controllori."]} +{"source": "Deep learning on graphs has become a popular research topic with many applications. However, past work has concentrated on learning graph embedding tasks only, which is in contrast with advances in generative models for images and text. Is it possible to transfer this progress to the domain of graphs? We propose to sidestep hurdles associated with linearization of such discrete structures by having a decoder output a probabilistic fully-connected graph of a predefined maximum size directly at once. Our method is formulated as a variational autoencoder. We evaluate on the challenging task of conditional molecule generation.", "target": ["Dimostriamo un autoencoder per i grafi.", "Imparare a generare grafi usando metodi di deep learning in \"one shot\", producendo direttamente le probabilità di esistenza di nodi e archi e i vettori di attributi dei nodi.", "Un encoder automatico variazionale per generare grafi"]} +{"source": "Long Short-Term Memory (LSTM) is one of the most widely used recurrent structures in sequence modeling. Its goal is to use gates to control the information flow (e.g., whether to skip some information/transformation or not) in the recurrent computations, although its practical implementation based on soft gates only partially achieves this goal and is easy to overfit. In this paper, we propose a new way for LSTM training, which pushes the values of the gates towards 0 or 1. By doing so, we can (1) better control the information flow: the gates are mostly open or closed, instead of in a middle state; and (2) avoid overfitting to certain extent: the gates operate at their flat regions, which is shown to correspond to better generalization ability. However, learning towards discrete values of the gates is generally difficult. To tackle this challenge, we leverage the recently developed Gumbel-Softmax trick from the field of variational methods, and make the model trainable with standard backpropagation. Experimental results on language modeling and machine translation show that (1) the values of the gates generated by our method are more reasonable and intuitively interpretable, and (2) our proposed method generalizes better and achieves better accuracy on test sets in all tasks. Moreover, the learnt models are not sensitive to low-precision approximation and low-rank approximation of the gate parameters due to the flat loss surface.", "target": ["Proponiamo un nuovo algoritmo per il training LSTM forzando il learning verso gate a valore binario che abbiamo dimostrato avere molte proprietà desiderabili.", "Proporre una nuova funzione \"gate\" per LSTM per abilitare i valori dei gate verso 0 o 1. ", "L'articolo mira a spingere le porte LSTM ad essere binarie impiegando il recente trucco del Gumbel-Softmax per ottenere una distribuzione categorica addestrabile end-to-end."]} +{"source": "We present a personalized recommender system using neural network for recommending products, such as eBooks, audio-books, Mobile Apps, Video and Music. It produces recommendations based on customer’s implicit feedback history such as purchases, listens or watches. Our key contribution is to formulate recommendation problem as a model that encodes historical behavior to predict the future behavior using soft data split, combining predictor and auto-encoder models. We introduce convolutional layer for learning the importance (time decay) of the purchases depending on their purchase date and demonstrate that the shape of the time decay function can be well approximated by a parametrical function. We present offline experimental results showing that neural networks with two hidden layers can capture seasonality changes, and at the same time outperform other modeling techniques, including our recommender in production. Most importantly, we demonstrate that our model can be scaled to all digital categories, and we observe significant improvements in an online A/B test. We also discuss key enhancements to the neural network model and describe our production pipeline. Finally we open-sourced our deep learning library which supports multi-gpu model parallel training. This is an important feature in building neural network based recommenders with large dimensionality of input and output data.", "target": ["Migliorare le raccomandazioni utilizzando la modellazione sensibile al tempo con reti neurali in più categorie di prodotti su un sito web di vendita al dettaglio", "L'articolo propone un nuovo metodo basato su reti neurali per la raccomandazione.", "Gli autori descrivono una procedura di costruzione del loro sistema di production recommender system da zero e integrano il decadimento temporale degli acquisti nel framework di apprendimento."]} +{"source": "Deep Learning (DL) algorithms based on Generative Adversarial Network (GAN) have demonstrated great potentials in computer vision tasks such as image restoration. Despite the rapid development of image restoration algorithms using DL and GANs, image restoration for specific scenarios, such as medical image enhancement and super-resolved identity recognition, are still facing challenges. How to ensure visually realistic restoration while avoiding hallucination or mode- collapse? How to make sure the visually plausible results do not contain hallucinated features jeopardizing downstream tasks such as pathology identification and subject identification? Here we propose to resolve these challenges by coupling the GAN based image restoration framework with another task-specific network. With medical imaging restoration as an example, the proposed model conducts additional pathology recognition/classification task to ensure the preservation of detailed structures that are important to this task. Validated on multiple medical datasets, we demonstrate the proposed method leads to improved deep learning based image restoration while preserving the detailed structure and diagnostic features. Additionally, the trained task network show potentials to achieve super-human level performance in identifying pathology and diagnosis. Further validation on super-resolved identity recognition tasks also show that the proposed method can be generalized for diverse image restoration tasks.", "target": ["Accoppiare il framework di restauro dell'immagine basata su GAN con un'altra rete specifica per il task per generare un'immagine realistica preservando le feature specifiche del task.", "Un nuovo metodo di Task-GAN di accoppiamento dell'immagine che accoppia GAN e una rete specifica del task, che aiuta ad evitare l'allucinazione o il mode collapse.", "Gli autori propongono di aumentare il restauro di immagini basato su GAN con un altro ramo specifico, come i task di classificazione, per un ulteriore miglioramento."]} +{"source": "Unsupervised anomaly detection on multi- or high-dimensional data is of great importance in both fundamental machine learning research and industrial applications, for which density estimation lies at the core. Although previous approaches based on dimensionality reduction followed by density estimation have made fruitful progress, they mainly suffer from decoupled model learning with inconsistent optimization goals and incapability of preserving essential information in the low-dimensional space. In this paper, we present a Deep Autoencoding Gaussian Mixture Model (DAGMM) for unsupervised anomaly detection. Our model utilizes a deep autoencoder to generate a low-dimensional representation and reconstruction error for each input data point, which is further fed into a Gaussian Mixture Model (GMM). Instead of using decoupled two-stage training and the standard Expectation-Maximization (EM) algorithm, DAGMM jointly optimizes the parameters of the deep autoencoder and the mixture model simultaneously in an end-to-end fashion, leveraging a separate estimation network to facilitate the parameter learning of the mixture model. The joint optimization, which well balances autoencoding reconstruction, density estimation of latent representation, and regularization, helps the autoencoder escape from less attractive local optima and further reduce reconstruction errors, avoiding the need of pre-training. Experimental results on several public benchmark datasets show that, DAGMM significantly outperforms state-of-the-art anomaly detection techniques, and achieves up to 14% improvement based on the standard F1 score.", "target": ["Una deep neural network addestrata end-to-end che sfrutta la Gaussian Mixture Modeling per eseguire la stima della densità e il rilevamento unsupervised delle anomalie in uno spazio bidimensionale appreso da un deep autoencoder.", "L'articolo presenta un framework di deep learning congiunto per la riduzione delle dimensioni e il clustering, che porta a un rilevamento delle anomalie con prestazioni competitive.", "Una nuova tecnica per il rilevamento delle anomalie in cui la riduzione delle dimensioni e la stima della densità sono ottimizzate congiuntamente."]} +{"source": "Generalization from limited examples, usually studied under the umbrella of meta-learning, equips learning techniques with the ability to adapt quickly in dynamical environments and proves to be an essential aspect of lifelong learning. In this paper, we introduce the Projective Subspace Networks (PSN), a deep learning paradigm that learns non-linear embeddings from limited supervision. In contrast to previous studies, the embedding in PSN deems samples of a given class to form an affine subspace. We will show that such modeling leads to robust solutions, yielding competitive results on supervised and semi-supervised few-shot classification. Moreover, our PSN approach has the ability of end-to-end learning. In contrast to previous works, our projective subspace can be thought of as a richer representation capturing higher-order information datapoints for modeling new concepts.", "target": ["Abbiamo proposto le reti a sottospazio proiettivo per few shot learning e semi-supervised few shot learning", "Questo articolo propone un nuovo approccio basato sull'embedding per il problema del few shot learning e un'estensione di questo modello al setting dell'apprendimento semi-supervised few shot.", "Nuovo metodo per la classificazione completamente e semi-supervisionata few shot basato sull'apprendimento di un'embedding generale e poi sull'apprendimento di un sottospazio di esso per ogni classe"]} +{"source": "This paper investigates whether learning contingency-awareness and controllable aspects of an environment can lead to better exploration in reinforcement learning. To investigate this question, we consider an instantiation of this hypothesis evaluated on the Arcade Learning Element (ALE). In this study, we develop an attentive dynamics model (ADM) that discovers controllable elements of the observations, which are often associated with the location of the character in Atari games. The ADM is trained in a self-supervised fashion to predict the actions taken by the agent. The learned contingency information is used as a part of the state representation for exploration purposes. We demonstrate that combining actor-critic algorithm with count-based exploration using our representation achieves impressive results on a set of notoriously challenging Atari games due to sparse rewards. For example, we report a state-of-the-art score of >11,000 points on Montezuma's Revenge without using expert demonstrations, explicit high-level information (e.g., RAM states), or supervisory data. Our experiments confirm that contingency-awareness is indeed an extremely powerful concept for tackling exploration problems in reinforcement learning and opens up interesting research questions for further investigations.", "target": ["Studiamo la consapevolezza della contingenza e gli aspetti controllabili nell'esplorazione e raggiungiamo prestazioni allo stato dell'arte su Montezuma's Revenge senza dimostrazioni di esperti.", "Questo articolo indaga il problema dell'estrazione di una rappresentazione di stato significativa per aiutare l'esplorazione quando ci si confronta con un task con reward sparsa, identificando le feature controllabili (apprese) dello stato", "Questo articolo propone l'idea nuova di usare la consapevolezza della contingenza per aiutare l'esplorazione in task di reinforcement learning a reward sparsa, ottenendo risultati allo stato dell'arte."]} +{"source": "Disentangling factors of variation has always been a challenging problem in representation learning. Existing algorithms suffer from many limitations, such as unpredictable disentangling factors, bad quality of generated images from encodings, lack of identity information, etc. In this paper, we proposed a supervised algorithm called DNA-GAN trying to disentangle different attributes of images. The latent representations of images are DNA-like, in which each individual piece represents an independent factor of variation. By annihilating the recessive piece and swapping a certain piece of two latent representations, we obtain another two different representations which could be decoded into images. In order to obtain realistic images and also disentangled representations, we introduced the discriminator for adversarial training. Experiments on Multi-PIE and CelebA datasets demonstrate the effectiveness of our method and the advantage of overcoming limitations existing in other methods.", "target": ["Abbiamo proposto un algoritmo supervisionato, DNA-GAN, per distinguere più attributi delle immagini.", "Questo articolo indaga il problema della generazione di immagini condizionata dagli attributi usando generative adversarial network, e propone di generare immagini da attributi e codice latente come rappresentazione di alto livello.", "Questo articolo ha proposto un nuovo metodo per distinguere diversi attributi delle immagini utilizzando un nuovo framework di DNA GAN"]} +{"source": "Representations learnt through deep neural networks tend to be highly informative, but opaque in terms of what information they learn to encode. We introduce an approach to probabilistic modelling that learns to represent data with two separate deep representations: an invariant representation that encodes the information of the class from which the data belongs, and an equivariant representation that encodes the symmetry transformation defining the particular data point within the class manifold (equivariant in the sense that the representation varies naturally with symmetry transformations). This approach to representation learning is conceptually transparent, easy to implement, and in-principle generally applicable to any data comprised of discrete classes of continuous distributions (e.g. objects in images, topics in language, individuals in behavioural data). We demonstrate qualitatively compelling representation learning and competitive quantitative performance, in both supervised and semi-supervised settings, versus comparable modelling approaches in the literature with little fine tuning.", "target": ["Questo articolo presenta una nuova tecnica di modellazione generativa di variabili latenti che permette la rappresentazione di informazioni globali in una variabile latente e di informazioni locali in un'altra variabile latente.", "L'articolo presenta un VAE che usa le label per separare la rappresentazione appresa in una parte invariante e una covariante."]} +{"source": "Convolutional neural networks (CNNs) have been successfully applied to many recognition and learning tasks using a universal recipe; training a deep model on a very large dataset of supervised examples. However, this approach is rather restrictive in practice since collecting a large set of labeled images is very expensive. One way to ease this problem is coming up with smart ways for choosing images to be labelled from a very large collection (i.e. active learning). Our empirical study suggests that many of the active learning heuristics in the literature are not effective when applied to CNNs when applied in batch setting. Inspired by these limitations, we define the problem of active learning as core-set selection, i.e. choosing set of points such that a model learned over the selected subset is competitive for the remaining data points. We further present a theoretical result characterizing the performance of any selected subset using the geometry of the datapoints. As an active learning algorithm, we choose the subset which is expected to yield best result according to our characterization. Our experiments show that the proposed method significantly outperforms existing approaches in image classification experiments by a large margin.", "target": ["Ci avviciniamo al problema dell'active learning come un problema di selezione del core-set e mostriamo che questo approccio è particolarmente utile nel setting del batch active learning che è cruciale quando si addestrano le CNN.", "Gli autori forniscono un algoritmo di active learning indipendente dall'algoritmo per la classificazione multiclasse", "L'articolo propone un algoritmo di active learning in modalità batch per CNN come un problema di core-set che supera il sampling casuale e il sampling di incertezza.", "Studia l'active learning per le reti neurali convoluzionali e formula il problema dell'active learning come selezione di core-set e presenta una nuova strategia"]} +{"source": "Recurrent neural networks are known for their notorious exploding and vanishing gradient problem (EVGP). This problem becomes more evident in tasks where the information needed to correctly solve them exist over long time scales, because EVGP prevents important gradient components from being back-propagated adequately over a large number of steps. We introduce a simple stochastic algorithm (\\textit{h}-detach) that is specific to LSTM optimization and targeted towards addressing this problem. Specifically, we show that when the LSTM weights are large, the gradient components through the linear path (cell state) in the LSTM computational graph get suppressed. Based on the hypothesis that these components carry information about long term dependencies (which we show empirically), their suppression can prevent LSTMs from capturing them. Our algorithm\\footnote{Our code is available at https://github.com/bhargav104/h-detach. } prevents gradients flowing through this path from getting suppressed, thus allowing the LSTM to capture such dependencies better. We show significant improvements over vanilla LSTM gradient based training in terms of convergence speed, robustness to seed and learning rate, and generalization using our modification of LSTM gradient on various benchmark datasets.", "target": ["Un semplice algoritmo per migliorare l'ottimizzazione e la gestione delle dipendenze a lungo termine in LSTM", "L'articolo introduce un semplice algoritmo stocastico chiamato h-detach che è specifico per l'ottimizzazione LSTM e mirato ad affrontare questo problema.", "Propone una semplice modifica al processo di training dell'LSTM per facilitare la propagazione del gradiente lungo gli stati delle cellule, o il \"percorso temporale lineare\"."]} +{"source": "Convolutional Neural Networks (CNNs) significantly improve the state-of-the-art for many applications, especially in computer vision. However, CNNs still suffer from a tendency to confidently classify out-distribution samples from unknown classes into pre-defined known classes. Further, they are also vulnerable to adversarial examples. We are relating these two issues through the tendency of CNNs to over-generalize for areas of the input space not covered well by the training set. We show that a CNN augmented with an extra output class can act as a simple yet effective end-to-end model for controlling over-generalization. As an appropriate training set for the extra class, we introduce two resources that are computationally efficient to obtain: a representative natural out-distribution set and interpolated in-distribution samples. To help select a representative natural out-distribution set among available ones, we propose a simple measurement to assess an out-distribution set's fitness. We also demonstrate that training such an augmented CNN with representative out-distribution natural datasets and some interpolated samples allows it to better handle a wide range of unseen out-distribution samples and black-box adversarial examples without training it on any adversaries. Finally, we show that generation of white-box adversarial attacks using our proposed augmented CNN can become harder, as the attack algorithms have to get around the rejection regions when generating actual adversaries.", "target": ["Addestrare correttamente le CNN con la classe dustbin aumenta la loro robustezza agli adversarial attack e la loro capacità di gestire i sample fuori distribuzione.", "Questo articolo propone di aggiungere una label aggiuntiva per rilevare i sample OOD e gli adversarial example nei modelli CNN.", "L'articolo propone una classe aggiuntiva che incorpora immagini naturali out-distribution e immagini interpolate per gli adversarial e out-distribution sample CNN"]} +{"source": "Modern deep artificial neural networks have achieved impressive results through models with very large capacity---compared to the number of training examples---that control overfitting with the help of different forms of regularization. Regularization can be implicit, as is the case of stochastic gradient descent or parameter sharing in convolutional layers, or explicit. Most common explicit regularization techniques, such as dropout and weight decay, reduce the effective capacity of the model and typically require the use of deeper and wider architectures to compensate for the reduced capacity. Although these techniques have been proven successful in terms of results, they seem to waste capacity. In contrast, data augmentation techniques reduce the generalization error by increasing the number of training examples and without reducing the effective capacity. In this paper we systematically analyze the effect of data augmentation on some popular architectures and conclude that data augmentation alone---without any other explicit regularization techniques---can achieve the same performance or higher as regularized models, especially when training with fewer examples.", "target": ["In una deep convolutional neural network con un livello sufficiente di data augmentation, ottimizzata da SGD, i regolarizzatori espliciti (weight decay e dropout) potrebbero non fornire alcun miglioramento aggiuntivo della generalizzazione.", "Questo articolo propone data augmentation come alternativa alle tecniche di regolarizzazione comunemente usate, e mostra che per alcuni modelli/task di riferimento la stessa performance di generalizzazione può essere ottenuta usando solo data augmentation.", "Questo articolo presenta uno studio sistematico della data augmentation nella classificazione delle immagini con deep neural network, suggerendo che data augmentation possa replicare alcuni regolarizzatori comuni come il weight decay e il dropout."]} +{"source": "Text editing on mobile devices can be a tedious process. To perform various editing operations, a user must repeatedly move his or her fingers between the text input area and the keyboard, making multiple round trips and breaking the flow of typing. In this work, we present Gedit, a system of on-keyboard gestures for convenient mobile text editing. Our design includes a ring gesture and flicks for cursor control, bezel gestures for mode switching, and four gesture shortcuts for copy, paste, cut, and undo. Variations of our gestures exist for one and two hands. We conducted an experiment to compare Gedit with the de facto touch+widget based editing interactions. Our results showed that Gedit’s gestures were easy to learn, 24% and 17% faster than the de facto interactions for one- and two-handed use, respectively, and preferred by participants.", "target": ["In questo lavoro, presentiamo Gedit, un sistema di gesture sulla tastiera per un comodo editing di testo su mobile.", "Riporta la progettazione e la valutazione delle tecniche di interazione di Gedit.", "Presenta una nuova serie di gesture tattili per eseguire la transizione senza soluzione di continuità tra l'inserimento e la modifica del testo nei dispositivi mobili"]} +{"source": "Deep learning achieves remarkable generalization capability with overwhelming number of model parameters. Theoretical understanding of deep learning generalization receives recent attention yet remains not fully explored. This paper attempts to provide an alternative understanding from the perspective of maximum entropy. We first derive two feature conditions that softmax regression strictly apply maximum entropy principle. DNN is then regarded as approximating the feature conditions with multilayer feature learning, and proved to be a recursive solution towards maximum entropy principle. The connection between DNN and maximum entropy well explains why typical designs such as shortcut and regularization improves model generalization, and provides instructions for future model development.", "target": ["Dimostriamo che una DNN è una soluzione approssimata ricorsivamente al principio di massima entropia.", "Presenta una derivazione che collega una DNN all'applicazione ricorsiva dell'adattamento del modello di massima entropia.", "L'articolo mira a fornire una visione del deep learning dalla prospettiva del principio di massima entropia."]} +{"source": "As people learn to navigate the world, autonomic nervous system (e.g., ``fight or flight) responses provide intrinsic feedback about the potential consequence of action choices (e.g., becoming nervous when close to a cliff edge or driving fast around a bend.) Physiological changes are correlated with these biological preparations to protect one-self from danger. We present a novel approach to reinforcement learning that leverages a task-independent intrinsic reward function trained on peripheral pulse measurements that are correlated with human autonomic nervous system responses. Our hypothesis is that such reward functions can circumvent the challenges associated with sparse and skewed rewards in reinforcement learning settings and can help improve sample efficiency. We test this in a simulated driving environment and show that it can increase the speed of learning and reduce the number of collisions during the learning stage.", "target": ["Presentiamo un nuovo approccio al reinforcement learning che sfrutta una funzione di reward intrinseca indipendente dal task, addestrata sulle misurazioni delle pulsazioni periferiche che sono correlate alle risposte del sistema nervoso autonomo dell'essere umano.", "Propone un framework di reinforcement learning basato sulla reazione emotiva umana nel contesto della guida autonoma.", "Gli autori propongono di utilizzare i segnali, come le risposte viscerali autonome di base che influenzano il processo decisionale, all'interno del framework di RL aumentando le funzioni di reward RL con un modello appreso direttamente dalle risposte del sistema nervoso umano.", "Propone di utilizzare i segnali fisiologici per migliorare le prestazioni degli algoritmi di reinforcement learning e costruire una funzione di reward intrinseca che è meno sparsa misurando l'ampiezza delle pulsazioni cardiache"]} +{"source": "Deep convolutional neural networks (CNNs) are known to be robust against label noise on extensive datasets. However, at the same time, CNNs are capable of memorizing all labels even if they are random, which means they can memorize corrupted labels. Are CNNs robust or fragile to label noise? Much of researches focusing on such memorization uses class-independent label noise to simulate label corruption, but this setting is simple and unrealistic. In this paper, we investigate the behavior of CNNs under class-dependently simulated label noise, which is generated based on the conceptual distance between classes of a large dataset (i.e., ImageNet-1k). Contrary to previous knowledge, we reveal CNNs are more robust to such class-dependent label noise than class-independent label noise. We also demonstrate the networks under class-dependent noise situations learn similar representation to the no noise situation, compared to class-independent noise situations.", "target": ["Le CNN sono robuste o fragili al rumore delle label? Nella pratica, robuste.", "Gli autori sfidano la robustezza delle CNN al rumore delle label usando l'albero ImageNet 1k di WordNet.", "Un'analisi delle prestazioni del modello di rete neurale convoluzionale quando viene introdotto rumore dipendente dalla classe e indipendente dalla classe", "Dimostra che le CNN sono più robuste al rumore delle label relativo alla classe e sostiene che il rumore del mondo reale dovrebbe essere relativo alla classe "]} +{"source": "Efficient audio synthesis is an inherently difficult machine learning task, as human perception is sensitive to both global structure and fine-scale waveform coherence. Autoregressive models, such as WaveNet, model local structure at the expense of global latent structure and slow iterative sampling, while Generative Adversarial Networks (GANs), have global latent conditioning and efficient parallel sampling, but struggle to generate locally-coherent audio waveforms. Herein, we demonstrate that GANs can in fact generate high-fidelity and locally-coherent audio by modeling log magnitudes and instantaneous frequencies with sufficient frequency resolution in the spectral domain. Through extensive empirical investigations on the NSynth dataset, we demonstrate that GANs are able to outperform strong WaveNet baselines on automated and human evaluation metrics, and efficiently generate audio several orders of magnitude faster than their autoregressive counterparts.", "target": ["Sintesi audio di alta qualità con le GAN", "Propone un approccio che utilizza il framework delle GAN per generare l'audio attraverso la modellazione di log magnitudini e frequenze istantanee con una risoluzione di frequenza sufficiente nel dominio spettrale.", "Una strategia per generare sample audio dal rumore con le GAN, con modifiche all'architettura e alla rappresentazione necessarie per generare un audio convincente che contenga un codice latente interpretabile.", "Presenta una semplice idea per rappresentare meglio i dati audio in modo da poter applicare modelli convoluzionali come le generative adversarial network"]} +{"source": "In this work we propose a novel approach for learning graph representation of the data using gradients obtained via backpropagation. Next we build a neural network architecture compatible with our optimization approach and motivated by graph filtering in the vertex domain. We demonstrate that the learned graph has richer structure than often used nearest neighbors graphs constructed based on features similarity. Our experiments demonstrate that we can improve prediction quality for several convolution on graphs architectures, while others appeared to be insensitive to the input graph.", "target": ["Ottimizzazione dei grafi con filtraggio del segnale nel dominio dei vertici.", "L'articolo studia l'apprendimento della matrice di adiacenza di un grafo con un grafo indiretto scarsamente connesso con pesi non negativi sugli archi e utilizza un algoritmo di projected sub-gradient descent.", "Sviluppa un nuovo schema per fare backpropagation sulla matrice di adiacenza di un grafo di una rete neurale"]} +{"source": "The use of AR in an industrial context could help for the training of new operators. To be able to use an AR guidance system, we need a tool to quickly create a 3D representation of the assembly line and of its AR annotations. This tool should be very easy to use by an operator who is not an AR or VR specialist: typically the manager of the assembly line. This is why we proposed WAAT, a 3D authoring tool allowing user to quickly create 3D models of the workstations, and also test the AR guidance placement. WAAT makes on-site authoring possible, which should really help to have an accurate 3D representation of the assembly line. The verification of AR guidance should also be very useful to make sure everything is visible and doesn't interfere with technical tasks. In addition to these features, our future work will be directed in the deployment of WAAT into a real boiler assembly line to assess the usability of this solution.", "target": ["Questo articolo descrive uno strumento di authoring 3D per fornire AR nelle linee di assemblaggio dell'industria 4.0", "L'articolo affronta il modo in cui gli strumenti di authoring AR supportino il training dei sistemi della catena di montaggio e propone un approccio", "Un sistema di guida AR per linee di assemblaggio industriali che permette l'authoring in loco di contenuti AR.", "Presenta un sistema che permette agli operai di una fabbrica di essere addestrati in modo più efficiente usando un sistema di realtà aumentata."]} +{"source": "Generative adversarial network (GAN) is one of the best known unsupervised learning techniques these days due to its superior ability to learn data distributions. In spite of its great success in applications, GAN is known to be notoriously hard to train. The tremendous amount of time it takes to run the training algorithm and its sensitivity to hyper-parameter tuning have been haunting researchers in this area. To resolve these issues, we need to first understand how GANs work. Herein, we take a step toward this direction by examining the dynamics of GANs. We relate a large class of GANs including the Wasserstein GANs to max-min optimization problems with the coupling term being linear over the discriminator. By developing new primal-dual optimization tools, we show that, with a proper stepsize choice, the widely used first-order iterative algorithm in training GANs would in fact converge to a stationary solution with a sublinear rate. The same framework also applies to multi-task learning and distributional robust learning problems. We verify our analysis on numerical examples with both synthetic and real data sets. We hope our analysis shed light on future studies on the theoretical properties of relevant machine learning problems.", "target": ["Mostriamo che, con una scelta appropriata della stepsize, l'algoritmo iterativo del primo ordine ampiamente utilizzato nel training delle GAN convergerebbe a una soluzione stazionaria con un tasso sublineare.", "Questo articolo usa le GAN e multi-task learning per fornire una garanzia di convergenza per algoritmi primali-duali su certi problemi min-max.", "Analizza la dinamica di apprendimento delle GAN formulando il problema come un problema di ottimizzazione primale-duale assumendo una classe limitata di modelli"]} +{"source": "Social dilemmas, where mutual cooperation can lead to high payoffs but participants face incentives to cheat, are ubiquitous in multi-agent interaction. We wish to construct agents that cooperate with pure cooperators, avoid exploitation by pure defectors, and incentivize cooperation from the rest. However, often the actions taken by a partner are (partially) unobserved or the consequences of individual actions are hard to predict. We show that in a large class of games good strategies can be constructed by conditioning one's behavior solely on outcomes (ie. one's past rewards). We call this consequentialist conditional cooperation. We show how to construct such strategies using deep reinforcement learning techniques and demonstrate, both analytically and experimentally, that they are effective in social dilemmas beyond simple matrix games. We also show the limitations of relying purely on consequences and discuss the need for understanding both the consequences of and the intentions behind an action.", "target": ["Mostriamo come usare deep RL per costruire agenti che possono risolvere dilemmi sociali al di là dei matrix game.", "Imparare a giocare giochi a somma generale a due giocatori con uno stato avente informazioni imperfette", "Specifica una strategia di innesco (CCC) e l'algoritmo corrispondente, dimostrando la convergenza verso risultati efficienti nei dilemmi sociali senza bisogno che gli agenti osservino le azioni degli altri."]} +{"source": "In distributed training, the communication cost due to the transmission of gradients or the parameters of the deep model is a major bottleneck in scaling up the number of processing nodes. To address this issue, we propose dithered quantization for the transmission of the stochastic gradients and show that training with Dithered Quantized Stochastic Gradients (DQSG) is similar to the training with unquantized SGs perturbed by an independent bounded uniform noise, in contrast to the other quantization methods where the perturbation depends on the gradients and hence, complicating the convergence analysis. We study the convergence of training algorithms using DQSG and the trade off between the number of quantization levels and the training time. Next, we observe that there is a correlation among the SGs computed by workers that can be utilized to further reduce the communication overhead without any performance loss. Hence, we develop a simple yet effective quantization scheme, nested dithered quantized SG (NDQSG), that can reduce the communication significantly without requiring the workers communicating extra information to each other. We prove that although NDQSG requires significantly less bits, it can achieve the same quantization variance bound as DQSG. Our simulation results confirm the effectiveness of training using DQSG and NDQSG in reducing the communication bits or the convergence time compared to the existing methods without sacrificing the accuracy of the trained model.", "target": ["L'articolo propone e analizza due schemi di quantizzazione per comunicare i gradienti stocastici nell'apprendimento distribuito che ridurrebbero i costi di comunicazione rispetto allo stato dell'arte, mantenendo la stessa precisione.", "Gli autori propongono di applicare la quantizzazione dithered ai gradienti stocastici calcolati durante il processo di training, che migliora l'errore di quantizzazione e raggiunge risultati superiori rispetto alle baseline, e propongono uno schema annidato per ridurre i costi di comunicazione.", "Gli autori stabiliscono una connessione tra la riduzione della comunicazione nell'ottimizzazione distribuita e la quantizzazione dithered e sviluppano due nuovi algoritmi di training distribuiti in cui l'overhead della comunicazione è significativamente ridotto."]} +{"source": "Deep neural networks have been shown to perform well in many classical machine learning problems, especially in image classification tasks. However, researchers have found that neural networks can be easily fooled, and they are surprisingly sensitive to small perturbations imperceptible to humans. Carefully crafted input images (adversarial examples) can force a well-trained neural network to provide arbitrary outputs. Including adversarial examples during training is a popular defense mechanism against adversarial attacks. In this paper we propose a new defensive mechanism under the generative adversarial network~(GAN) framework. We model the adversarial noise using a generative network, trained jointly with a classification discriminative network as a minimax game. We show empirically that our adversarial network approach works well against black box attacks, with performance on par with state-of-art methods such as ensemble adversarial training and adversarial training with projected gradient descent.", "target": ["Addestrare congiuntamente una rete generatrice di rumore adversarial con una rete di classificazione per fornire una migliore robustezza agli adversarial attack.", "Una soluzione GAN per deep model per la classificazione, di fronte agli attacchi white e black box, che produce modelli robusti.", "L'articolo propone un meccanismo difensivo contro gli adversarial attack utilizzando GAN con perturbazioni generate e poi usate come adversarial example e un discriminatore usato per distinguerli"]} +{"source": "Deep learning has become the state of the art approach in many machine learning problems such as classification. It has recently been shown that deep learning is highly vulnerable to adversarial perturbations. Taking the camera systems of self-driving cars as an example, small adversarial perturbations can cause the system to make errors in important tasks, such as classifying traffic signs or detecting pedestrians. Hence, in order to use deep learning without safety concerns a proper defense strategy is required. We propose to use ensemble methods as a defense strategy against adversarial perturbations. We find that an attack leading one model to misclassify does not imply the same for other networks performing the same task. This makes ensemble methods an attractive defense strategy against adversarial attacks. We empirically show for the MNIST and the CIFAR-10 data sets that ensemble methods not only improve the accuracy of neural networks on test data but also increase their robustness against adversarial perturbations.", "target": ["Utilizzo di metodi d'ensemble come difesa dalle adversarial perturbation prodotte contro le deep neural network.", "Questo articolo propone di utilizzare l'ensembling come un meccanismo di adversarial defense.", "Ha studiato empiricamente la robustezza di diversi ensemble di deep neural network ai due tipi di attacchi, FGSM e BIM, su due dataset popolari, MNIST e CIFAR10"]} +{"source": "In this paper, we propose the Associative Conversation Model that generates visual information from textual information and uses it for generating sentences in order to utilize visual information in a dialogue system without image input. In research on Neural Machine Translation, there are studies that generate translated sentences using both images and sentences, and these studies show that visual information improves translation performance. However, it is not possible to use sentence generation algorithms using images for the dialogue systems since many text-based dialogue systems only accept text input. Our approach generates (associates) visual information from input text and generates response text using context vector fusing associative visual information and sentence textual information. A comparative experiment between our proposed model and a model without association showed that our proposed model is generating useful sentences by associating visual information related to sentences. Furthermore, analysis experiment of visual association showed that our proposed model generates (associates) visual information effective for sentence generation.", "target": ["Proposta di un metodo di generazione di frasi basato sulla fusione tra informazioni testuali e informazioni visive associate alle informazioni testuali", "Questo lavoro descrive un modello di deep learning per sistemi di dialogo che sfrutta le informazioni visive.", "Questo articolo propone un nuovo dataset per il grounded dialog e fa un'osservazione computazionale che potrebbe aiutare a ragionare sulla visione anche quando si esegue un dialogo basato sul testo.", "Propone di arricchire i tradizionali approcci di generazione di frasi/dialogo basati sul testo incorporando informazioni visive, raccogliendo un dataset che consiste sia di testo che di immagini o video associati"]} +{"source": "Feedforward convolutional neural network has achieved a great success in many computer vision tasks. While it validly imitates the hierarchical structure of biological visual system, it still lacks one essential architectural feature: contextual recurrent connections with feedback, which widely exists in biological visual system. In this work, we designed a Contextual Recurrent Convolutional Network with this feature embedded in a standard CNN structure. We found that such feedback connections could enable lower layers to ``rethink\" about their representations given the top-down contextual information. We carefully studied the components of this network, and showed its robustness and superiority over feedforward baselines in such tasks as noise image classification, partially occluded object recognition and fine-grained image classification. We believed this work could be an important step to help bridge the gap between computer vision models and real biological visual system.", "target": ["Abbiamo proposto una nuova rete convoluzionale ricorrente contestuale con proprietà robuste di apprendimento visivo", "Questo articolo introduce la connessione di feedback per migliorare l'apprendimento delle feature attraverso l'embedding di informazioni del contesto.", "L'articolo propone di aggiungere connessioni \"ricorrenti\" all'interno di una rete di convoluzione con meccanismo di gating."]} +{"source": "Deep neural networks have led to a series of breakthroughs, dramatically improving the state-of-the-art in many domains. The techniques driving these advances, however, lack a formal method to account for model uncertainty. While the Bayesian approach to learning provides a solid theoretical framework to handle uncertainty, inference in Bayesian-inspired deep neural networks is difficult. In this paper, we provide a practical approach to Bayesian learning that relies on a regularization technique found in nearly every modern network, batch normalization. We show that training a deep network using batch normalization is equivalent to approximate inference in Bayesian models, and we demonstrate how this finding allows us to make useful estimates of the model uncertainty. Using our approach, it is possible to make meaningful uncertainty estimates using conventional architectures without modifying the network or the training procedure. Our approach is thoroughly validated in a series of empirical experiments on different tasks and using various measures, showing it to outperform baselines on a majority of datasets with strong statistical significance.", "target": ["Mostriamo che il training di una deep network usando la batch normalization è equivalente all'inferenza approssimativa nei modelli bayesiani, e dimostriamo come questa scoperta ci permette di fare stime utili dell'incertezza del modello nelle reti convenzionali.", "Questo articolo propone di usare la batch normalization al momento del test per ottenere l'incertezza predittiva, e mostra che la predizione Monte Carlo al momento del test usando batch norm è migliore del dropout.", "Propone che la procedura di regolarizzazione chiamata batch normalization possa essere intesa come l'esecuzione di un'inferenza bayesiana approssimata, che si comporta in modo simile al dropout MC in termini delle stime di incertezza che produce."]} +{"source": "Data-parallel neural network training is network-intensive, so gradient dropping was designed to exchange only large gradients. However, gradient dropping has been shown to slow convergence. We propose to improve convergence by having each node combine its locally computed gradient with the sparse global gradient exchanged over the network. We empirically confirm with machine translation tasks that gradient dropping with local gradients approaches convergence 48% faster than non-compressed multi-node training and 28% faster compared to vanilla gradient dropping. We also show that gradient dropping with a local gradient update does not reduce the model's final quality.", "target": ["Miglioriamo il gradient dropping (una tecnica per scambiare solo gradienti di grandi dimensioni nel training distribuito) incorporando i gradienti locali mentre facciamo un aggiornamento dei parametri per ridurre la loss di qualità e migliorare ulteriormente il tempo di training.", "Questo articolo propone 3 modalità per combinare i gradienti locali e globali per utilizzare meglio più nodi di calcolo", "Si occupa del problema della riduzione dei requisiti di comunicazione per l'implementazione delle tecniche di ottimizzazione distribuita, in particolare SGD"]} +{"source": "We establish the relation between Distributional RL and the Upper Confidence Bound (UCB) approach to exploration. In this paper we show that the density of the Q function estimated by Distributional RL can be successfully used for the estimation of UCB. This approach does not require counting and, therefore, generalizes well to the Deep RL. We also point to the asymmetry of the empirical densities estimated by the Distributional RL algorithms like QR-DQN. This observation leads to the reexamination of the variance's performance in the UCB type approach to exploration. We introduce truncated variance as an alternative estimator of the UCB and a novel algorithm based on it. We empirically show that newly introduced algorithm achieves better performance in multi-armed bandits setting. Finally, we extend this approach to high-dimensional setting and test it on the Atari 2600 games. New approach achieves better performance compared to QR-DQN in 26 of games, 13 ties out of 49 games.", "target": ["Esplorazione con RL distributivo e varianza troncata.", "Presenta un metodo RL per gestire i trade-off di exploration-exploitation tramite tecniche UCB.", "Un metodo per utilizzare la distribuzione appresa dalla Quantile Regression DQN per l'esplorazione, al posto della solita strategia epsilon-greedy.", "Propone nuovi algoritmi (QUCB e QUCB+) per gestire il tradeoff di esplorazione in Multi-Armed Bandit e più in generale nel Reinforcement Learning"]} +{"source": "Good representations facilitate transfer learning and few-shot learning. Motivated by theories of language and communication that explain why communities with large number of speakers have, on average, simpler languages with more regularity, we cast the representation learning problem in terms of learning to communicate. Our starting point sees traditional autoencoders as a single encoder with a fixed decoder partner that must learn to communicate. Generalizing from there, we introduce community-based autoencoders in which multiple encoders and decoders collectively learn representations by being randomly paired up on successive training iterations. Our experiments show that increasing community sizes reduce idiosyncrasies in the learned codes, resulting in more invariant representations with increased reusability and structure.", "target": ["Motivati dalle teorie del linguaggio e della comunicazione, introduciamo autoencoder basati sulla comunità, in cui encoder e decoder multipli imparano collettivamente rappresentazioni strutturate e riutilizzabili.", "Gli autori affrontano il problema del representation learning, mirano a costruire una rappresentazione riutilizzabile e strutturata, sostengono che il co-adattamento tra encoder e decoder nell'AE tradizionale produce una scarsa rappresentazione, e introducono autoencoder basati sulla comunità.", "L'articolo presenta un framework di autoencoder basato sulla comunità per affrontare il co-adattamento di codificatori e decodificatori e mira a costruire rappresentazioni migliori."]} +{"source": "Humans are experts at high-fidelity imitation -- closely mimicking a demonstration, often in one attempt. Humans use this ability to quickly solve a task instance, and to bootstrap learning of new tasks. Achieving these abilities in autonomous agents is an open problem. In this paper, we introduce an off-policy RL algorithm (MetaMimic) to narrow this gap. MetaMimic can learn both (i) policies for high-fidelity one-shot imitation of diverse novel skills, and (ii) policies that enable the agent to solve tasks more efficiently than the demonstrators. MetaMimic relies on the principle of storing all experiences in a memory and replaying these to learn massive deep neural network policies by off-policy RL. This paper introduces, to the best of our knowledge, the largest existing neural networks for deep RL and shows that larger networks with normalization are needed to achieve one-shot high-fidelity imitation on a challenging manipulation task. The results also show that both types of policy can be learned from vision, in spite of the task rewards being sparse, and without access to demonstrator actions.", "target": ["Presentiamo MetaMimic, un algoritmo che prende come input un dataset di dimostrazioni e produce (i) una policy di imitazione ad alta fedeltà one-shot e (ii) una policy di task incondizionata.", "L'articolo esamina il problema dell'imitazione one-shot con alta precisione di imitazione, estendendo DDPGfD per usare solo traiettorie di stato.", "Questo articolo propone un approccio per l'imitazione one-shot con alta precisione, e affronta il problema comune dell'esplorazione nell'imitation learning.", "Presenta un metodo di RL per l'apprendimento dalla dimostrazione video senza accesso alle azioni degli esperti"]} +{"source": "Normalization methods are a central building block in the deep learning toolbox. They accelerate and stabilize training, while decreasing the dependence on manually tuned learning rate schedules. When learning from multi-modal distributions, the effectiveness of batch normalization (BN), arguably the most prominent normalization method, is reduced. As a remedy, we propose a more flexible approach: by extending the normalization to more than a single mean and variance, we detect modes of data on-the-fly, jointly normalizing samples that share common features. We demonstrate that our method outperforms BN and other widely used normalization techniques in several experiments, including single and multi-task datasets.", "target": ["Presentiamo un nuovo metodo di normalizzazione per le deep neural network che è robusto alle multi-modalità nelle distribuzioni di feature intermedie.", "Metodo di normalizzazione che impara la distribuzione multi-modale nello spazio delle feature", "Propone una generalizzazione della Batch Normalization sotto l'ipotesi che la statistica delle attivazioni unitarie sulle batch e sulle dimensioni spaziali non sia unimodale"]} +{"source": "Multilingual machine translation, which translates multiple languages with a single model, has attracted much attention due to its efficiency of offline training and online serving. However, traditional multilingual translation usually yields inferior accuracy compared with the counterpart using individual models for each language pair, due to language diversity and model capacity limitations. In this paper, we propose a distillation-based approach to boost the accuracy of multilingual machine translation. Specifically, individual models are first trained and regarded as teachers, and then the multilingual model is trained to fit the training data and match the outputs of individual models simultaneously through knowledge distillation. Experiments on IWSLT, WMT and Ted talk translation datasets demonstrate the effectiveness of our method. Particularly, we show that one model is enough to handle multiple languages (up to 44 languages in our experiment), with comparable or even better accuracy than individual models.", "target": ["Abbiamo proposto un metodo basato sulla knowledge distillation per aumentare la precisione della traduzione automatica neurale multilingue.", "Un modello di traduzione automatica multilingue many-to-one che prima addestra modelli separati per ogni coppia di lingue e poi esegue la distillation.", "L'articolo mira al training di un modello di traduzione automatica aumentando la loss standard di cross-entropy con una componente di distillation basata su teacher model individuali (single-language-pair)."]} +{"source": "What makes humans so good at solving seemingly complex video games? Unlike computers, humans bring in a great deal of prior knowledge about the world, enabling efficient decision making. This paper investigates the role of human priors for solving video games. Given a sample game, we conduct a series of ablation studies to quantify the importance of various priors. We do this by modifying the video game environment to systematically mask different types of visual information that could be used by humans as priors. We find that removal of some prior knowledge causes a drastic degradation in the speed with which human players solve the game, e.g. from 2 minutes to over 20 minutes. Furthermore, our results indicate that general priors, such as the importance of objects and visual consistency, are critical for efficient game-play.", "target": ["Indaghiamo i vari tipi di conoscenza preliminare che aiutano l'apprendimento umano e troviamo che i prior generali sugli oggetti giocano il ruolo più critico nel guidare il gioco umano.", "Gli autori studiano tramite un esperimento quali aspetti dei prior umani sono importanti per reinforcement learning nei videogiochi.", "Gli autori presentano uno studio sui prior impiegati dagli esseri umani nel giocare ai videogiochi e dimostrano l'esistenza di una tassonomia di feature che influenzano la capacità di completare i task nel gioco in vari gradi."]} +{"source": "Driven by the need for parallelizable hyperparameter optimization methods, this paper studies \\emph{open loop} search methods: sequences that are predetermined and can be generated before a single configuration is evaluated. Examples include grid search, uniform random search, low discrepancy sequences, and other sampling distributions. In particular, we propose the use of $k$-determinantal point processes in hyperparameter optimization via random search. Compared to conventional uniform random search where hyperparameter settings are sampled independently, a $k$-DPP promotes diversity. We describe an approach that transforms hyperparameter search spaces for efficient use with a $k$-DPP. In addition, we introduce a novel Metropolis-Hastings algorithm which can sample from $k$-DPPs defined over spaces with a mixture of discrete and continuous dimensions. Our experiments show significant benefits over uniform random search in realistic scenarios with a limited budget for training supervised learners, whether in serial or parallel.", "target": ["Spinti dalla necessità di metodi di ottimizzazione degli iperparametri parallelizzabili e open-loop, proponiamo l'uso di k-determinantal point process nell'ottimizzazione degli iperparametri tramite ricerca casuale.", "Propone di usare il k-DPP per selezionare i punti candidati nelle ricerche di iperparametri.", "Gli autori propongono k-DPP come un metodo open-loop per l'ottimizzazione degli iperparametri e forniscono il suo studio empirico e il confronto con altri metodi.", "Considera la ricerca degli iperparametri non sequenziale e non informata usando determinantal point process, che sono distribuzioni di probabilità su sottoinsiemi di un ground set con la proprietà che i sottoinsiemi con elementi più 'diversi' hanno una probabilità più alta"]} +{"source": "In inductive transfer learning, fine-tuning pre-trained convolutional networks substantially outperforms training from scratch. When using fine-tuning, the underlying assumption is that the pre-trained model extracts generic features, which are at least partially relevant for solving the target task, but would be difficult to extract from the limited amount of data available on the target task. However, besides the initialization with the pre-trained model and the early stopping, there is no mechanism in fine-tuning for retaining the features learned on the source task. In this paper, we investigate several regularization schemes that explicitly promote the similarity of the final solution with the initial model. We eventually recommend a simple $L^2$ penalty using the pre-trained model as a reference, and we show that this approach behaves much better than the standard scheme using weight decay on a partially frozen network.", "target": ["Nel transfer learning induttivo, il fine-tuning di reti convoluzionali pre-addestrate supera sostanzialmente il training da zero.", "Affronta il problema del transfer learning nelle deep network e propone di avere un termine di regolarizzazione che penalizza la divergenza dall'inizializzazione.", "Propone un'analisi su diverse tecniche di regolarizzazione adattiva per il deep transfer learning, concentrandosi in particolare sull'uso di una condizione L@-SP"]} +{"source": "Artificial neural networks have opened up a world of possibilities in data science and artificial intelligence, but neural networks are cumbersome tools that grow with the complexity of the learning problem. We make contributions to this issue by considering a modified version of the fully connected layer we call a block diagonal inner product layer. These modified layers have weight matrices that are block diagonal, turning a single fully connected layer into a set of densely connected neuron groups. This idea is a natural extension of group, or depthwise separable, convolutional layers applied to the fully connected layers. Block diagonal inner product layers can be achieved by either initializing a purely block diagonal weight matrix or by iteratively pruning off diagonal block entries. This method condenses network storage and speeds up the run time without significant adverse effect on the testing accuracy, thus offering a new approach to improve network computation efficiency.", "target": ["Studiamo le reti neurali con layer di prodotti interni diagonali a blocchi per l'efficienza.", "Questo articolo propone che i layer interni di una rete neurale siano diagonali a blocchi, e discute sul fatto che le matrici diagonali a blocchi siano più efficienti del pruning e che i layer diagonali a blocchi portino a reti più efficienti.", "Sostituzione di layer fully connected con layer fully connected block-diagonal"]} +{"source": "One of the challenges in the study of generative adversarial networks is the instability of its training. In this paper, we propose a novel weight normalization technique called spectral normalization to stabilize the training of the discriminator. Our new normalization technique is computationally light and easy to incorporate into existing implementations. We tested the efficacy of spectral normalization on CIFAR10, STL-10, and ILSVRC2012 dataset, and we experimentally confirmed that spectrally normalized GANs (SN-GANs) is capable of generating images of better or equal quality relative to the previous training stabilization techniques.", "target": ["Proponiamo una nuova tecnica di normalizzazione dei pesi chiamata normalizzazione spettrale per stabilizzare il training del discriminatore delle GAN.", "Questo articolo usa la regolarizzazione spettrale per normalizzare gli obiettivi GAN, e la GAN che ne risulta, chiamato SN-GAN, assicura essenzialmente la proprietà di Lipschitz del discriminatore.", "Questo articolo propone la \"normalizzazione spettrale\", facendo un bel passo in avanti nel miglioramento del training delle GAN."]} +{"source": "Humans acquire complex skills by exploiting previously learned skills and making transitions between them. To empower machines with this ability, we propose a method that can learn transition policies which effectively connect primitive skills to perform sequential tasks without handcrafted rewards. To efficiently train our transition policies, we introduce proximity predictors which induce rewards gauging proximity to suitable initial states for the next skill. The proposed method is evaluated on a set of complex continuous control tasks in bipedal locomotion and robotic arm manipulation which traditional policy gradient methods struggle at. We demonstrate that transition policies enable us to effectively compose complex skills with existing primitive skills. The proposed induced rewards computed using the proximity predictor further improve training efficiency by providing more dense information than the sparse rewards from the environments. We make our environments, primitive skills, and code public for further research at https://youngwoon.github.io/transition .", "target": ["Le policy di transizione permettono agli agenti di comporre skill complesse collegando in modo fluido le skill primitive precedentemente acquisite.", "Propone uno schema per la transizione a stati di strating favorevoli per l'esecuzione di opzioni date in domini continui. Utilizza due processi di apprendimento eseguiti simultaneamente.", "Presenta un metodo per l'apprendimento delle policy di transizione da un task ad un altro con l'obiettivo di completare task complessi utilizzando lo stimatore di prossimità di stato per premiare la policy di transizione.", "Propone un nuovo schema di training con una funzione di reward ausiliaria appresa per ottimizzare le policy di transizione che collegano lo stato finale di una precedente macro azione/opzione con buoni stati iniziali della seguente macro azione/opzione"]} +{"source": "Gated recurrent units (GRUs) were inspired by the common gated recurrent unit, long short-term memory (LSTM), as a means of capturing temporal structure with less complex memory unit architecture. Despite their incredible success in tasks such as natural and artificial language processing, speech, video, and polyphonic music, very little is understood about the specific dynamic features representable in a GRU network. As a result, it is difficult to know a priori how successful a GRU-RNN will perform on a given data set. In this paper, we develop a new theoretical framework to analyze one and two dimensional GRUs as a continuous dynamical system, and classify the dynamical features obtainable with such system. We found rich repertoire that includes stable limit cycles over time (nonlinear oscillations), multi-stable state transitions with various topologies, and homoclinic orbits. In addition, we show that any finite dimensional GRU cannot precisely replicate the dynamics of a ring attractor, or more generally, any continuous attractor, and is limited to finitely many isolated fixed points in theory. These findings were then experimentally verified in two dimensions by means of time series prediction.", "target": ["Classifichiamo le feature dinamiche che una e due celle GRU possono e non possono catturare in tempo continuo, e verifichiamo sperimentalmente i nostri risultati con la predizione di serie temporali in k-step.", "Gli autori analizzano le GRU con hidden size di uno e due come sistemi dinamici a tempo continuo, sostenendo che la potenza espressiva della rappresentazione degli hidden state può fornire una conoscenza preliminare su quanto bene una GRU si comporterà su un dato dataset", "Questo articolo analizza le GRU da una prospettiva di sistemi dinamici, e mostra che le GRU 2d possono essere addestrate ad adottare una varietà di punti fissi e possono approssimare gli attrattori di linea, ma non possono imitare un attrattore ad anello.", "Converte le equazioni di GRU in tempo continuo e usa la teoria e gli esperimenti per studiare reti GRU a 1 e 2 dimensioni e mostrare ogni varietà di topologia dinamica disponibile in questi sistemi"]} +{"source": "Stacked hourglass network has become an important model for Human pose estimation. The estimation of human body posture depends on the global information of the keypoints type and the local information of the keypoints location. The consistent processing of inputs and constraints makes it difficult to form differentiated and determined collaboration mechanisms for each stacked hourglass network. In this paper, we propose a Multi-Scale Stacked Hourglass (MSSH) network to high-light the differentiation capabilities of each Hourglass network for human pose estimation. The pre-processing network forms feature maps of different scales,and dispatch them to various locations of the stack hourglass network, where the small-scale features reach the front of stacked hourglass network, and large-scale features reach the rear of stacked hourglass network. And a new loss function is proposed for multi-scale stacked hourglass network. Different keypoints have different weight coefficients of loss function at different scales, and the keypoints weight coefficients are dynamically adjusted from the top-level hourglass network to the bottom-level hourglass network. Experimental results show that the pro-posed method is competitive with respect to the comparison algorithm on MPII and LSP datasets.", "target": ["Gli input differenziati causano una differenziazione funzionale della rete, e l'interazione delle funzioni di loss tra le reti può influenzare il processo di ottimizzazione.", "Una modifica alla rete originale a clessidra per la single pose estimation che produce miglioramenti rispetto alla baseline originale.", "Gli autori estendono una stacked hourglass network con moduli inception-resnet-A e propongono un approccio multiscala per la stima della posa umana in immagini RGB fisse."]} +{"source": "We present a new unsupervised method for learning general-purpose sentence embeddings. Unlike existing methods which rely on local contexts, such as words inside the sentence or immediately neighboring sentences, our method selects, for each target sentence, influential sentences in the entire document based on a document structure. We identify a dependency structure of sentences using metadata or text styles. Furthermore, we propose a novel out-of-vocabulary word handling technique to model many domain-specific terms, which were mostly discarded by existing sentence embedding methods. We validate our model on several tasks showing 30% precision improvement in coreference resolution in a technical domain, and 7.5% accuracy increase in paraphrase detection compared to baselines.", "target": ["Per addestrare un modello di sentence embedding utilizzando documenti tecnici, il nostro approccio considera la struttura del documento per trovare un contesto più ampio e gestire le parole fuori dal vocabolario.", "Presenta idee per migliorare i sentence embedding attingendo a più contesti.", "Apprendimento di rappresentazioni di frasi con informazioni sulle dipendenze delle frasi", "Estende l'idea di formare una rappresentazione non supervisionata delle frasi usata nell'approccio SkipThough utilizzando un insieme più ampio di prove per formare la rappresentazione di una frase"]} +{"source": "Neural network training relies on our ability to find ````````\"good\" minimizers of highly non-convex loss functions. It is well known that certain network architecture designs (e.g., skip connections) produce loss functions that train easier, and well-chosen training parameters (batch size, learning rate, optimizer) produce minimizers that generalize better. However, the reasons for these differences, and their effect on the underlying loss landscape, is not well understood. In this paper, we explore the structure of neural loss functions, and the effect of loss landscapes on generalization, using a range of visualization methods. First, we introduce a simple ``\"filter normalization\" method that helps us visualize loss function curvature, and make meaningful side-by-side comparisons between loss functions. Then, using a variety of visualizations, we explore how network architecture effects the loss landscape, and how training parameters affect the shape of minimizers.", "target": ["Esploriamo la struttura delle funzioni di loss neurali e l'effetto dei landscape delle loss sulla generalizzazione, usando una serie di metodi di visualizzazione.", "Questo articolo propone un metodo per visualizzare la loss di una NN e fornisce approfondimenti sull'addestrabilità e la generalizzazione delle NN.", "Studia la non convessità della superficie di loss e dei percorsi di ottimizzazione."]} +{"source": "Deep models are state-of-the-art for many computer vision tasks including image classification and object detection. However, it has been shown that deep models are vulnerable to adversarial examples. We highlight how one-hot encoding directly contributes to this vulnerability and propose breaking away from this widely-used, but highly-vulnerable mapping. We demonstrate that by leveraging a different output encoding, multi-way encoding, we can make models more robust. Our approach makes it more difficult for adversaries to find useful gradients for generating adversarial attacks. We present state-of-the-art robustness results for black-box, white-box attacks, and achieve higher clean accuracy on four benchmark datasets: MNIST, CIFAR-10, CIFAR-100, and SVHN when combined with adversarial training. The strength of our approach is also presented in the form of an attack for model watermarking, raising challenges in detecting stolen models.", "target": ["Dimostriamo che sfruttando una codifica dell'output a più vie, piuttosto che la codifica one-hot ampiamente utilizzata, possiamo rendere i deep model più robusti agli adversarial attack.", "Questo articolo propone di sostituire il layer finale di cross-entropy addestrato sulle label one-hot nei classificatori codificando ogni label come un vettore ad alta densità e addestrando il classificatore per minimizzare la distanza L2 dalla codifica della classe corretta.", "Gli autori propongono un nuovo metodo contro gli adversarial attack che mostra una quantità significativa di guadagni rispetto alle baseline"]} +{"source": "Existing approaches to neural machine translation condition each output word on previously generated outputs. We introduce a model that avoids this autoregressive property and produces its outputs in parallel, allowing an order of magnitude lower latency during inference. Through knowledge distillation, the use of input token fertilities as a latent variable, and policy gradient fine-tuning, we achieve this at a cost of as little as 2.0 BLEU points relative to the autoregressive Transformer network used as a teacher. We demonstrate substantial cumulative improvements associated with each of the three aspects of our training strategy, and validate our approach on IWSLT 2016 English–German and two WMT language pairs. By sampling fertilities in parallel at inference time, our non-autoregressive model achieves near-state-of-the-art performance of 29.8 BLEU on WMT 2016 English–Romanian.", "target": ["Introduciamo il primo modello NMT con decodifica completamente parallela, riducendo la latenza di inferenza di 10 volte.", "Questo lavoro propone un decoder non autoregressivo per il framework encoder-decoder in cui la decisione di generare una parola non dipende dalle decisioni precedenti sulle parole generate", "Questo articolo descrive un approccio alla decodifica non autoregressiva per la traduzione automatica neurale con la possibilità di una decodifica più parallela che può risultare in un significativo aumento della velocità.", "Propone l'introduzione di un insieme di variabili latenti per rappresentare la fertilità di ogni parola sorgente per rendere la generazione della frase target non autoregressiva"]} +{"source": "While neural networks have achieved high accuracy on standard image classification benchmarks, their accuracy drops to nearly zero in the presence of small adversarial perturbations to test inputs. Defenses based on regularization and adversarial training have been proposed, but often followed by new, stronger attacks that defeat these defenses. Can we somehow end this arms race? In this work, we study this problem for neural networks with one hidden layer. We first propose a method based on a semidefinite relaxation that outputs a certificate that for a given network and test input, no attack can force the error to exceed a certain value. Second, as this certificate is differentiable, we jointly optimize it with the network parameters, providing an adaptive regularizer that encourages robustness against all attacks. On MNIST, our approach produces a network and a certificate that no that perturbs each pixel by at most $\\epsilon = 0.1$ can cause more than $35\\%$ test error.", "target": ["Dimostriamo un metodo certificabile, addestrabile e scalabile per difendersi dagli adversarial example.", "Propone una nuova difesa contro gli attacchi di sicurezza alle reti neurali con il modello attaccante che produce un certificato di sicurezza sull'algoritmo.", "Deriva un limite superiore all'adversarial perturbation per le reti neurali con un hidden layer "]} +{"source": "We formulate an information-based optimization problem for supervised classification. For invertible neural networks, the control of these information terms is passed down to the latent features and parameter matrix in the last fully connected layer, given that mutual information is invariant under invertible map. We propose an objective function and prove that it solves the optimization problem. Our framework allows us to learn latent features in an more interpretable form while improving the classification performance. We perform extensive quantitative and qualitative experiments in comparison with the existing state-of-the-art classification models.", "target": ["Proponiamo un regolarizzatore che migliora le prestazioni di classificazione delle reti neurali", "Gli autori propongono di addestrare un modello da un punto di massimizzazione dell'informazione reciproca tra le predizioni e gli output veri, con un termine di regolarizzazione che minimizza l'informazione irrilevante durante l'apprendimento.", "Propone di decomporre i parametri in una mappa di feature invertibile F e una trasformazione lineare w nell'ultimo layer per massimizzare l'informazione reciproca I(Y, \\hat{T}) vincolando l'informazione irrilevante"]} +{"source": "Powerful generative models, particularly in Natural Language Modelling, are commonly trained by maximizing a variational lower bound on the data log likelihood. These models often suffer from poor use of their latent variable, with ad-hoc annealing factors used to encourage retention of information in the latent variable. We discuss an alternative and general approach to latent variable modelling, based on an objective that encourages a perfect reconstruction by tying a stochastic autoencoder with a variational autoencoder (VAE). This ensures by design that the latent variable captures information about the observations, whilst retaining the ability to generate well. Interestingly, although our model is fundamentally different to a VAE, the lower bound attained is identical to the standard VAE bound but with the addition of a simple pre-factor; thus, providing a formal interpretation of the commonly used, ad-hoc pre-factors in training VAEs.", "target": ["Questo articolo introduce un nuovo framework di modellazione generativa che evita il collasso delle variabili latenti e chiarisce l'uso di alcuni fattori ad-hoc nel training degli autoencoder variazionali.", "L'articolo propone di risolvere il problema di un auto-encoder variazionale che ignora le variabili latenti.", "Questo articolo propone di aggiungere un autoencoder stocastico al modello VAE originale per affrontare il problema che il decoder LSTM di un language model potrebbe essere troppo forte per ignorare le informazioni della variabile latente.", "Questo articolo presenta AutoGen, che combina un autoencoder variazionale generativo con un modello di ricostruzione ad alta fedeltà basato su autoencoder per utilizzare meglio la rappresentazione latente"]} +{"source": "This paper studies the problem of domain division which aims to segment instances drawn from different probabilistic distributions. This problem exists in many previous recognition tasks, such as Open Set Learning (OSL) and Generalized Zero-Shot Learning (G-ZSL), where the testing instances come from either seen or unseen/novel classes with different probabilistic distributions. Previous works only calibrate the confident prediction of classifiers of seen classes (WSVM Scheirer et al. (2014)) or taking unseen classes as outliers Socher et al. (2013). In contrast, this paper proposes a probabilistic way of directly estimating and fine-tuning the decision boundary between seen and unseen classes. In particular, we propose a domain division algorithm to split the testing instances into known, unknown and uncertain domains, and then conduct recognition tasks in each domain. Two statistical tools, namely, bootstrapping and KolmogorovSmirnov (K-S) Test, for the first time, are introduced to uncover and fine-tune the decision boundary of each domain. Critically, the uncertain domain is newly introduced in our framework to adopt those instances whose domain labels cannot be predicted confidently. Extensive experiments demonstrate that our approach achieved the state-of-the-art performance on OSL and G-ZSL benchmarks.", "target": ["Questo articolo studia il problema della divisione del dominio segmentando le istanze tratte da diverse distribuzioni di probabilità.", "Questo articolo si occupa del problema del riconoscimento della novità nell'open set learning e nello zero-shot learning generalizzato e propone una possibile soluzione", "Un approccio alla separazione dei domini basato sul bootstrapping per identificare le soglie di cutoff di somiglianza per le classi note, seguito da un test di Kolmogorov-Smirnoff per raffinare le zone in-distribution bootstrapped.", "Propone di introdurre un nuovo dominio, il dominio incerto, per gestire meglio la divisione tra domini visti/non visti nell'open set learning e nel zero-shot learning generalizzato"]} +{"source": "Stochastic gradient descent (SGD) is widely believed to perform implicit regularization when used to train deep neural networks, but the precise manner in which this occurs has thus far been elusive. We prove that SGD minimizes an average potential over the posterior distribution of weights along with an entropic regularization term. This potential is however not the original loss function in general. So SGD does perform variational inference, but for a different loss than the one used to compute the gradients. Even more surprisingly, SGD does not even converge in the classical sense: we show that the most likely trajectories of SGD for deep networks do not behave like Brownian motion around critical points. Instead, they resemble closed loops with deterministic components. We prove that such out-of-equilibrium behavior is a consequence of highly non-isotropic gradient noise in SGD; the covariance matrix of mini-batch gradients for deep networks has a rank as small as 1% of its dimension. We provide extensive empirical validation of these claims, proven in the appendix.", "target": ["SGD esegue implicitamente l'inferenza variazionale; il rumore del gradiente è altamente non isotropo, quindi SGD non converge nemmeno ai punti critici della loss originale", "Questo articolo fornisce un'analisi variazionale di SGD come processo di non-equilibrio.", "Questo articolo discute la funzione obiettivo regolarizzata minimizzata da SGD standard nel contesto delle reti neurali, e fornisce una prospettiva di inferenza variazionale utilizzando l'equazione di Fokker-Planck.", "Sviluppa una teoria per studiare l'impatto del rumore del gradiente stocastico per SGD, specialmente per i modelli di deep neural network"]} +{"source": "The current dominant paradigm for imitation learning relies on strong supervision of expert actions to learn both 'what' and 'how' to imitate. We pursue an alternative paradigm wherein an agent first explores the world without any expert supervision and then distills its experience into a goal-conditioned skill policy with a novel forward consistency loss. In our framework, the role of the expert is only to communicate the goals (i.e., what to imitate) during inference. The learned policy is then employed to mimic the expert (i.e., how to imitate) after seeing just a sequence of images demonstrating the desired task. Our method is 'zero-shot' in the sense that the agent never has access to expert actions during training or for the task demonstration at inference. We evaluate our zero-shot imitator in two real-world settings: complex rope manipulation with a Baxter robot and navigation in previously unseen office environments with a TurtleBot. Through further experiments in VizDoom simulation, we provide evidence that better mechanisms for exploration lead to learning a more capable policy which in turn improves end task performance. Videos, models, and more details are available at https://pathak22.github.io/zeroshot-imitation/.", "target": ["Gli agenti possono imparare a imitare solo dimostrazioni visive (senza azioni) al momento del test dopo aver imparato dalla propria esperienza senza alcuna forma di supervisione al momento del training.", "Questo articolo propone un approccio per visual learning zero-shot tramite l'apprendimento di skill function parametriche.", "Un articolo sull'imitazione di un task presentato proprio durante l'inferenza, dove l'apprendimento viene eseguito in modo self-supervised e durante il training l'agente esplora task correlati ma diversi.", "Propone un metodo per aggirare il problema della costose dimostrazioni da parte di esperti utilizzando l'esplorazione casuale di un agente per imparare abilità generalizzabili che possono essere applicate senza uno specifico pretraining"]} +{"source": "Distributional Semantics Models(DSM) derive word space from linguistic items in context. Meaning is obtained by defining a distance measure between vectors corresponding to lexical entities. Such vectors present several problems. This work concentrates on quality of word embeddings, improvement of word embedding vectors, applicability of a novel similarity metric used ‘on top’ of the word embeddings. In this paper we provide comparison between two methods for post process improvements to the baseline DSM vectors. The counter-fitting method which enforces antonymy and synonymy constraints into the Paragram vector space representations recently showed improvement in the vectors’ capability for judging semantic similarity. The second method is our novel RESM method applied to GloVe baseline vectors. By applying the hubness reduction method, implementing relational knowledge into the model by retrofitting synonyms and providing a new ranking similarity definition RESM that gives maximum weight to the top vector component values we equal the results for the ESL and TOEFL sets in comparison with our calculations using the Paragram and Paragram + Counter-fitting methods. For SIMLEX-999 gold standard since we cannot use the RESM the results using GloVe and PPDB are significantly worse compared to Paragram. Apparently, counter-fitting corrects hubness. The Paragram or our cosine retrofitting method are state-of-the-art results for the SIMLEX-999 gold standard. They are 0.2 better for SIMLEX-999 than word2vec with sense de-conflation (that was announced to be state-of the-art method for less reliable gold standards). Apparently relational knowledge and counter-fitting is more important for judging semantic similarity than sense determination for words. It is to be mentioned, though that Paragram hyperparameters are fitted to SIMLEX-999 results. The lesson is that many corrections to word embeddings are necessary and methods with more parameters and hyperparameters perform better.", "target": ["L'articolo fornisce una descrizione di una procedura per migliorare il modello di spazio vettoriale delle parole con una valutazione dei modelli Paragram e GloVe per i benchmark di similarità.", "Questo articolo suggerisce un nuovo algoritmo che regola i word vector di GloVe e poi usa una funzione di somiglianza non euclidea tra loro.", "Gli autori presentano osservazioni sulle debolezze dei modelli di spazio vettoriale esistenti ed elencano un approccio in 6 fasi per raffinare i word vector esistenti"]} +{"source": "Recurrent neural networks have achieved excellent performance in many applications. However, on portable devices with limited resources, the models are often too large to deploy. For applications on the server with large scale concurrent requests, the latency during inference can also be very critical for costly computing resources. In this work, we address these problems by quantizing the network, both weights and activations, into multiple binary codes {-1,+1}. We formulate the quantization as an optimization problem. Under the key observation that once the quantization coefficients are fixed the binary codes can be derived efficiently by binary search tree, alternating minimization is then applied. We test the quantization for two well-known RNNs, i.e., long short term memory (LSTM) and gated recurrent unit (GRU), on the language models. Compared with the full-precision counter part, by 2-bit quantization we can achieve ~16x memory saving and ~6x real inference acceleration on CPUs, with only a reasonable loss in the accuracy. By 3-bit quantization, we can achieve almost no loss in the accuracy or even surpass the original model, with ~10.5x memory saving and ~3x real inference acceleration. Both results beat the exiting quantization works with large margins. We extend our alternating quantization to image classification tasks. In both RNNs and feedforward neural networks, the method also achieves excellent performance.", "target": ["Proponiamo un nuovo metodo di quantizzazione e lo applichiamo per quantizzare le RNN sia per la compressione che per l'accelerazione", "Questo articolo propone un metodo di quantizzazione multi-bit per reti neurali ricorrenti.", "Una tecnica per quantizzare le matrici di peso delle reti neurali e una procedura di ottimizzazione alternata per stimare l'insieme di k vettori binari e coefficienti che rappresentino meglio il vettore originale."]} +{"source": "The goal of this paper is to demonstrate a method for tensorizing neural networks based upon an efficient way of approximating scale invariant quantum states, the Multi-scale Entanglement Renormalization Ansatz (MERA). We employ MERA as a replacement for linear layers in a neural network and test this implementation on the CIFAR-10 dataset. The proposed method outperforms factorization using tensor trains, providing greater compression for the same level of accuracy and greater accuracy for the same level of compression. We demonstrate MERA-layers with 3900 times fewer parameters and a reduction in accuracy of less than 1% compared to the equivalent fully connected layers.", "target": ["Sostituiamo i layer fully connected di una rete neurale con il multi-scale entanglement renormalization ansatz, un tipo di operazione quantistica che descrive le correlazioni a lungo raggio.", "Nell'articolo gli autori suggeriscono di usare la tecnica di tensorizzazione MERA per comprimere le reti neurali.", "Una nuova parametrizzazione delle mappe lineari per l'uso nelle reti neurali, utilizzando una fattorizzazione gerarchica della mappa lineare che riduce il numero di parametri pur permettendo di modellare interazioni relativamente complesse.", "Studia come comprimere i layer feed forward usando decomposizioni tensoriali di basso rango ed esplorano una decomposizione ad albero"]} +{"source": "Deep learning models have outperformed traditional methods in many fields such as natural language processing and computer vision. However, despite their tremendous success, the methods of designing optimal Convolutional Neural Networks (CNNs) are still based on heuristics or grid search. The resulting networks obtained using these techniques are often overparametrized with huge computational and memory requirements. This paper focuses on a structured, explainable approach towards optimal model design that maximizes accuracy while keeping computational costs tractable. We propose a single-shot analysis of a trained CNN that uses Principal Component Analysis (PCA) to determine the number of filters that are doing significant transformations per layer, without the need for retraining. It can be interpreted as identifying the dimensionality of the hypothesis space under consideration. The proposed technique also helps estimate an optimal number of layers by looking at the expansion of dimensions as the model gets deeper. This analysis can be used to design an optimal structure of a given network on a dataset, or help to adapt a predesigned network on a new dataset. We demonstrate these techniques by optimizing VGG and AlexNet networks on CIFAR-10, CIFAR-100 and ImageNet datasets.", "target": ["Presentiamo un'analisi single shot di una rete neurale addestrata per rimuovere la ridondanza e identifichiamo la struttura ottimale della rete", "Questo articolo propone una serie di euristiche per identificare una buona architettura di rete neurale, basata sulla PCA delle attivazioni delle unità sul dataset", "Questo articolo presenta un framework per ottimizzare le architetture delle reti neurali attraverso l'identificazione dei filtri ridondanti tra i layer"]} +{"source": "Recent work has introduced attacks that extract the architecture information of deep neural networks (DNN), as this knowledge enhances an adversary’s capability to conduct attacks on black-box networks. This paper presents the first in-depth security analysis of DNN fingerprinting attacks that exploit cache side-channels. First, we define the threat model for these attacks: our adversary does not need the ability to query the victim model; instead, she runs a co-located process on the host machine victim ’s deep learning (DL) system is running and passively monitors the accesses of the target functions in the shared framework. Second, we introduce DeepRecon, an attack that reconstructs the architecture of the victim network by using the internal information extracted via Flush+Reload, a cache side-channel technique. Once the attacker observes function invocations that map directly to architecture attributes of the victim network, the attacker can reconstruct the victim’s entire network architecture. In our evaluation, we demonstrate that an attacker can accurately reconstruct two complex networks (VGG19 and ResNet50) having only observed one forward propagation. Based on the extracted architecture attributes, we also demonstrate that an attacker can build a meta-model that accurately fingerprints the architecture and family of the pre-trained model in a transfer learning setting. From this meta-model, we evaluate the importance of the observed attributes in the fingerprinting process. Third, we propose and evaluate new framework-level defense techniques that obfuscate our attacker’s observations. Our empirical security analysis represents a step toward understanding the DNNs’ vulnerability to cache side-channel attacks.", "target": ["Conduciamo la prima analisi di sicurezza approfondita degli attacchi di fingerprinting alle DNN che sfruttano i side-channel della cache, il che rappresenta un passo verso la comprensione delle vulnerabilità delle DNN agli attacchi side-channel.", "Questo articolo considera il problema del fingerprinting delle architetture di reti neurali utilizzando i side-channel della cache, e discute le difese security-through-obscurity.", "Questo articolo esegue attacchi cache side-channel per estrarre gli attributi di un modello vittima e dedurre la sua architettura, oltre a mostrare che possono raggiungere una precisione di classificazione quasi perfetta."]} +{"source": "Learning with a primary objective, such as softmax cross entropy for classification and sequence generation, has been the norm for training deep neural networks for years. Although being a widely-adopted approach, using cross entropy as the primary objective exploits mostly the information from the ground-truth class for maximizing data likelihood, and largely ignores information from the complement (incorrect) classes. We argue that, in addition to the primary objective, training also using a complement objective that leverages information from the complement classes can be effective in improving model performance. This motivates us to study a new training paradigm that maximizes the likelihood of the ground-truth class while neutralizing the probabilities of the complement classes. We conduct extensive experiments on multiple tasks ranging from computer vision to natural language understanding. The experimental results confirm that, compared to the conventional training with just one primary objective, training also with the complement objective further improves the performance of the state-of-the-art models across all tasks. In addition to the accuracy improvement, we also show that models trained with both primary and complement objectives are more robust to single-step adversarial attacks.", "target": ["Proponiamo il Complement Objective Training (COT), un nuovo paradigma di training che ottimizza sia gli obiettivi primari che quelli complementari per apprendere efficacemente i parametri delle reti neurali.", "Considera l'aggiunta all'obiettivo di cross-entropy della massimizzazione dell'obiettivo \"complemento\", che mira a neutralizzare le probabilità previste di classi diverse dalle label corretta.", "Gli autori propongono un obiettivo secondario per la minimizzazione del softmax basato sulla valutazione delle informazioni raccolte dalle classi errate, portando a un nuovo approccio di training.", "Si occupa del training di reti neurali per task di classificazione o generazione di sequenze utilizzando la cross-entropy loss"]} +{"source": "We present a new method for uncertainty estimation and out-of-distribution detection in neural networks with softmax output. We extend softmax layer with an additional constant input. The corresponding additional output is able to represent the uncertainty of the network. The proposed method requires neither additional parameters nor multiple forward passes nor input preprocessing nor out-of-distribution datasets. We show that our method performs comparably to more computationally expensive methods and outperforms baselines on our experiments from image recognition and sentiment analysis domains.", "target": ["Stima dell'incertezza in una singola forward pass senza parametri aggiuntivi apprendibili.", "Un nuovo metodo per calcolare le stime di incertezza dell'output nelle DNN per problemi di classificazione che compete coi metodi allo stato dell'arte per la stima dell'incertezza e li supera nei task di out-of-distribution detection.", "Gli autori presentano il softmax inibito, una modifica del softmax attraverso l'aggiunta di un'attivazione costante che fornisce una misura per l'incertezza."]} +{"source": "When deep learning is applied to sensitive data sets, many privacy-related implementation issues arise. These issues are especially evident in the healthcare, finance, law and government industries. Homomorphic encryption could allow a server to make inferences on inputs encrypted by a client, but to our best knowledge, there has been no complete implementation of common deep learning operations, for arbitrary model depths, using homomorphic encryption. This paper demonstrates a novel approach, efficiently implementing many deep learning functions with bootstrapped homomorphic encryption. As part of our implementation, we demonstrate Single and Multi-Layer Neural Networks, for the Wisconsin Breast Cancer dataset, as well as a Convolutional Neural Network for MNIST. Our results give promising directions for privacy-preserving representation learning, and the return of data control to users.", "target": ["Abbiamo realizzato un sistema ricco di funzionalità per il deep learning con input criptati, producendo output criptati, preservando la privacy.", "Un framework per l'inferenza privata del modello di deep learning utilizzando schemi FHE che supportano il bootstrapping veloce e quindi possono ridurre il tempo di calcolo.", "L'articolo presenta un mezzo per valutare una rete neurale in modo sicuro usando la crittografia omomorfa."]} +{"source": "In this paper, we introduce a system called GamePad that can be used to explore the application of machine learning methods to theorem proving in the Coq proof assistant. Interactive theorem provers such as Coq enable users to construct machine-checkable proofs in a step-by-step manner. Hence, they provide an opportunity to explore theorem proving with human supervision. We use GamePad to synthesize proofs for a simple algebraic rewrite problem and train baseline models for a formalization of the Feit-Thompson theorem. We address position evaluation (i.e., predict the number of proof steps left) and tactic prediction (i.e., predict the next proof step) tasks, which arise naturally in tactic-based theorem proving.", "target": ["Introduciamo un sistema chiamato GamePad per esplorare l'applicazione di metodi di apprendimento automatico alla dimostrazione di teoremi nell'assistente di prova Coq.", "Questo articolo descrive un sistema per applicare l'apprendimento automatico alla dimostrazione interattiva dei teoremi, si concentra sui task di predizione della tattica e di valutazione della posizione, e mostra che un modello neurale supera un SVM su entrambi i task.", "Propone che le tecniche di apprendimento automatico siano utilizzate per aiutare a costruire dimostrazioni nel theorem prover Coq."]} +{"source": "Deep neural networks are usually huge, which significantly limits the deployment on low-end devices. In recent years, many weight-quantized models have been proposed. They have small storage and fast inference, but training can still be time-consuming. This can be improved with distributed learning. To reduce the high communication cost due to worker-server synchronization, recently gradient quantization has also been proposed to train deep networks with full-precision weights. In this paper, we theoretically study how the combination of both weight and gradient quantization affects convergence. We show that (i) weight-quantized models converge to an error related to the weight quantization resolution and weight dimension; (ii) quantizing gradients slows convergence by a factor related to the gradient quantization resolution and dimension; and (iii) clipping the gradient before quantization renders this factor dimension-free, thus allowing the use of fewer bits for gradient quantization. Empirical experiments confirm the theoretical convergence results, and demonstrate that quantized networks can speed up training and have comparable performance as full-precision networks.", "target": ["In questo articolo, abbiamo studiato il training efficiente delle reti a peso quantizzato con gradiente quantizzato in un ambiente distribuito, sia teoricamente che empiricamente.", "Questo articolo studia le proprietà di convergenza della quantizzazione del peso consapevole della loss con diverse precisioni del gradiente nell'ambiente distribuito, e fornisce un'analisi di convergenza per la quantizzazione del peso con gradienti completi, quantizzati e quantizzati clipped.", "Gli autori propongono un'analisi dell'effetto della quantizzazione simultanea dei pesi e dei gradienti nel training di un modello parametrizzato in un ambiente distribuito completamente sincronizzato."]} +{"source": "Sequential learning, also called lifelong learning, studies the problem of learning tasks in a sequence with access restricted to only the data of the current task. In this paper we look at a scenario with fixed model capacity, and postulate that the learning process should not be selfish, i.e. it should account for future tasks to be added and thus leave enough capacity for them. To achieve Selfless Sequential Learning we study different regularization strategies and activation functions. We find that imposing sparsity at the level of the representation (i.e. neuron activations) is more beneficial for sequential learning than encouraging parameter sparsity. In particular, we propose a novel regularizer, that encourages representation sparsity by means of neural inhibition. It results in few active neurons which in turn leaves more free neurons to be utilized by upcoming tasks. As neural inhibition over an entire layer can be too drastic, especially for complex tasks requiring strong representations, our regularizer only inhibits other neurons in a local neighbourhood, inspired by lateral inhibition processes in the brain. We combine our novel regularizer with state-of-the-art lifelong learning methods that penalize changes to important previously learned parts of the network. We show that our new regularizer leads to increased sparsity which translates in consistent performance improvement on diverse datasets.", "target": ["Una strategia di regolarizzazione per migliorare le prestazioni del sequential learning", "Un nuovo approccio basato sulla regolarizzazione al problema del sequential learning utilizzando un modello a dimensione fissa che aggiunge termini extra alla loss, incoraggiando la sparsità della rappresentazione e combattendo il catastrophic forgetting.", "Questo articolo affronta il problema del catastrophic forgetting nel lifelong learning proponendo strategie di apprendimento regolarizzate"]} +{"source": "A Synaptic Neural Network (SynaNN) consists of synapses and neurons. Inspired by the synapse research of neuroscience, we built a synapse model with a nonlinear synapse function of excitatory and inhibitory channel probabilities. Introduced the concept of surprisal space and constructed a commutative diagram, we proved that the inhibitory probability function -log(1-exp(-x)) in surprisal space is the topologically conjugate function of the inhibitory complementary probability 1-x in probability space. Furthermore, we found that the derivative of the synapse over the parameter in the surprisal space is equal to the negative Bose-Einstein distribution. In addition, we constructed a fully connected synapse graph (tensor) as a synapse block of a synaptic neural network. Moreover, we proved the gradient formula of a cross-entropy loss function over parameters, so synapse learning can work with the gradient descent and backpropagation algorithms. In the proof-of-concept experiment, we performed an MNIST training and testing on the MLP model with synapse network as hidden layers.", "target": ["Una rete neurale sinaptica con grafo di sinapsi e un processo di learning che ha la feature della coniugazione topologica e della distribuzione di Bose-Einstein nello spazio di sorpresa.", "Gli autori propongono una rete neurale ibrida composta da un grafo di sinapsi che può essere incorporato in una rete neurale standard", "Presenta un modello di rete neurale biologicamente ispirato basato sui canali ionici eccitatori e inibitori nelle membrane delle cellule reali"]} +{"source": "Many types of relations in physical, biological, social and information systems can be modeled as homogeneous or heterogeneous concept graphs. Hence, learning from and with graph embeddings has drawn a great deal of research interest recently, but only ad hoc solutions have been obtained this far. In this paper, we conjecture that the one-shot supervised learning mechanism is a bottleneck in improving the performance of the graph embedding learning algorithms, and propose to extend this by introducing a multi-shot unsupervised learning framework. Empirical results on several real-world data set show that the proposed model consistently and significantly outperforms existing state-of-the-art approaches on knowledge base completion and graph based multi-label classification tasks.", "target": ["Modelli generalizzati di graph embedding", "Un approccio generalizzato di embedding dei knowledge graph che impara gli embedding basati su tre diversi obiettivi simultanei, e performa alla pari o addirittura supera gli approcci esistenti allo stato dell'arte.", "Affronta il task di apprendere embedding di grafi multi-relazionali usando una rete neurale", "Propone un nuovo metodo, GEN, per calcolare embedding di grafi multirelazionali, in particolare perché le cosiddette E-Cells e R-Cells possano rispondere a query della forma (h,r,?),(?r,t), e (h,?,t)"]} +{"source": "We introduce and study minimax curriculum learning (MCL), a new method for adaptively selecting a sequence of training subsets for a succession of stages in machine learning. The subsets are encouraged to be small and diverse early on, and then larger, harder, and allowably more homogeneous in later stages. At each stage, model weights and training sets are chosen by solving a joint continuous-discrete minimax optimization, whose objective is composed of a continuous loss (reflecting training set hardness) and a discrete submodular promoter of diversity for the chosen subset. MCL repeatedly solves a sequence of such optimizations with a schedule of increasing training set size and decreasing pressure on diversity encouragement. We reduce MCL to the minimization of a surrogate function handled by submodular maximization and continuous gradient methods. We show that MCL achieves better performance and, with a clustering trick, uses fewer labeled samples for both shallow and deep models while achieving the same performance. Our method involves repeatedly solving constrained submodular maximization of an only slowly varying function on the same ground set. Therefore, we develop a heuristic method that utilizes the previous submodular maximization solution as a warm start for the current submodular maximization process to reduce computation while still yielding a guarantee.", "target": ["Il Minimax Curriculum Learning è un metodo di machine learning che comporta l'aumento della hardness desiderabile e la riduzione programmata della diversità.", "Un approccio di curriculum learning che utilizza una funzione set submodulare che cattura la diversità degli esempi scelti durante il training.", "L'articolo introduce il MiniMax Curriculum learning come un approccio per il training adattivo dei modelli fornendo diversi sottoinsiemi di dati."]} +{"source": "Progress in probabilistic generative models has accelerated, developing richer models with neural architectures, implicit densities, and with scalable algorithms for their Bayesian inference. However, there has been limited progress in models that capture causal relationships, for example, how individual genetic factors cause major human diseases. In this work, we focus on two challenges in particular: How do we build richer causal models, which can capture highly nonlinear relationships and interactions between multiple causes? How do we adjust for latent confounders, which are variables influencing both cause and effect and which prevent learning of causal relationships? To address these challenges, we synthesize ideas from causality and modern probabilistic modeling. For the first, we describe implicit causal models, a class of causal models that leverages neural architectures with an implicit density. For the second, we describe an implicit causal model that adjusts for confounders by sharing strength across examples. In experiments, we scale Bayesian inference on up to a billion genetic measurements. We achieve state of the art accuracy for identifying causal factors: we significantly outperform the second best result by an absolute difference of 15-45.3%.", "target": ["Modelli impliciti applicati alla causalità e alla genetica", "Gli autori propongono di utilizzare il modello implicito per affrontare il problema dell'associazione Genome-Wide.", "Questo articolo propone soluzioni per i problemi di confondimento negli studi di associazione genome-wide dovuti alla struttura della popolazione e la potenziale presenza di interazioni non lineari tra diverse parti del genoma, e collega la genetica statistica e il ML.", "Presenta un modello generativo non lineare per GWAS che modella la struttura della popolazione in cui le non linearità sono modellate utilizzando reti neurali come approssimatori di funzioni non lineari e l'inferenza viene eseguita utilizzando l'inferenza variazionale senza likelihood"]} +{"source": "Few-shot learning trains image classifiers over datasets with few examples per category. It poses challenges for the optimization algorithms, which typically require many examples to fine-tune the model parameters for new categories. Distance-learning-based approaches avoid the optimization issue by embedding the images into a metric space and applying the nearest neighbor classifier for new categories. In this paper, we propose to exploit the object-level relation to learn the image relation feature, which is converted into a distance directly. For a new category, even though its images are not seen by the model, some objects may appear in the training images. Hence, object-level relation is useful for inferring the relation of images from unseen categories. Consequently, our model generalizes well for new categories without fine-tuning. Experimental results on benchmark datasets show that our approach outperforms state-of-the-art methods.", "target": ["Few-shot learning sfruttando la relazione object-level per imparare la relazione image-level (somiglianza)", "Questo articolo affronta il problema del few-shot learning proponendo un approccio basato sull'embedding che impara a confrontare le feature object-level tra gli example nel support e query set", "Propone un metodo di few-shot learning che sfrutta la relazione object-level tra diverse immagini basate sulla ricerca dei nearest neighbor e concatena le feature map di due immagini di input in un'unica feature map"]} +{"source": "Word embeddings are widely used in machine learning based natural language processing systems. It is common to use pre-trained word embeddings which provide benefits such as reduced training time and improved overall performance. There has been a recent interest in applying natural language processing techniques to programming languages. However, none of this recent work uses pre-trained embeddings on code tokens. Using extreme summarization as the downstream task, we show that using pre-trained embeddings on code tokens provides the same benefits as it does to natural languages, achieving: over 1.9x speedup, 5\\% improvement in test loss, 4\\% improvement in F1 scores, and resistance to over-fitting. We also show that the choice of language used for the embeddings does not have to match that of the task to achieve these benefits and that even embeddings pre-trained on human languages provide these benefits to programming languages.", "target": ["I ricercatori che esplorano le tecniche di natural language processing applicate al codice sorgente non stanno usando alcuna forma di embedding pre-addestrati, noi dimostriamo invece che dovrebbero farne uso.", "Questo articolo si prefigge di capire se il pretraining di word embedding per il codice nei linguaggi di programmazione utilizzando language model di NLP ha un impatto sul task di extreme code summarization.", "Questo lavoro mostra come il pre-training dei word vector usando corpora di codice porta a rappresentazioni che sono più adatte delle rappresentazioni randomly initialized e addestrate per la predizione dei nomi di funzioni/metodi"]} +{"source": "Recently, Approximate Policy Iteration (API) algorithms have achieved super-human proficiency in two-player zero-sum games such as Go, Chess, and Shogi without human data. These API algorithms iterate between two policies: a slow policy (tree search), and a fast policy (a neural network). In these two-player games, a reward is always received at the end of the game. However, the Rubik’s Cube has only a single solved state, and episodes are not guaranteed to terminate. This poses a major problem for these API algorithms since they rely on the reward received at the end of the game. We introduce Autodidactic Iteration: an API algorithm that overcomes the problem of sparse rewards by training on a distribution of states that allows the reward to propagate from the goal state to states farther away. Autodidactic Iteration is able to learn how to solve the Rubik’s Cube and the 15-puzzle without relying on human data. Our algorithm is able to solve 100% of randomly scrambled cubes while achieving a median solve length of 30 moves — less than or equal to solvers that employ human domain knowledge.", "target": ["Risolviamo il cubo di Rubik con il puro reinforcement learning", "Soluzione per risolvere il cubo di Rubik usando reinforcement learning (RL) con la ricerca ad albero di Monte-Carlo (MCTS) attraverso l'iterazione autodidattica.", "Questo lavoro risolve il cubo di Rubik usando un metodo di iterazione approssimativa delle policy chiamato iterazione autodidattica, superando il problema delle reward sparse creando il proprio sistema di reward.", "Introduce un algoritmo di deep RL per risolvere il cubo di Rubik che gestisce l'enorme spazio di stato e la reward molto sparsa del cubo di Rubik"]} +{"source": "Answering compositional questions requiring multi-step reasoning is challenging for current models. We introduce an end-to-end differentiable model for interpreting questions, which is inspired by formal approaches to semantics. Each span of text is represented by a denotation in a knowledge graph, together with a vector that captures ungrounded aspects of meaning. Learned composition modules recursively combine constituents, culminating in a grounding for the complete sentence which is an answer to the question. For example, to interpret ‘not green’, the model will represent ‘green’ as a set of entities, ‘not’ as a trainable ungrounded vector, and then use this vector to parametrize a composition function to perform a complement operation. For each sentence, we build a parse chart subsuming all possible parses, allowing the model to jointly learn both the composition operators and output structure by gradient descent. We show the model can learn to represent a variety of challenging semantic operators, such as quantifiers, negation, disjunctions and composed relations on a synthetic question answering task. The model also generalizes well to longer sentences than seen in its training data, in contrast to LSTM and RelNet baselines. We will release our code.", "target": ["Descriviamo un modello differenziabile end-to-end per QA che impara a rappresentare segmenti di testo nella domanda come denotazioni nel knowledge graph, imparando sia moduli neurali per la composizione che la struttura sintattica della frase.", "Questo articolo presenta un modello per la visual question answering che può imparare sia i parametri che gli structure predictor per una rete neurale modulare, senza strutture supervisionate o assistenza da un parser sintattico.", "Propone di addestrare un modello di question answering a partire dalle sole risposte e da una KB imparando alberi latenti che catturano la sintassi e imparano la semantica delle parole"]} +{"source": "Deep learning software demands reliability and performance. However, many of the existing deep learning frameworks are software libraries that act as an unsafe DSL in Python and a computation graph interpreter. We present DLVM, a design and implementation of a compiler infrastructure with a linear algebra intermediate representation, algorithmic differentiation by adjoint code generation, domain- specific optimizations and a code generator targeting GPU via LLVM. Designed as a modern compiler infrastructure inspired by LLVM, DLVM is more modular and more generic than existing deep learning compiler frameworks, and supports tensor DSLs with high expressivity. With our prototypical staged DSL embedded in Swift, we argue that the DLVM system enables a form of modular, safe and performant frameworks for deep learning.", "target": ["Introduciamo una nuova infrastruttura per un compilatore che affronta le carenze dei framework esistenti per il deep learning.", "Si propone di passare dalla generazione di codice ad-hoc nei motori di deep learning alle best practice in termini di compilatore e linguaggi di programmazione.", "Questo articolo presenta un framework per compilatori che permette la definizione di linguaggi specifici per i sistemi di deep learning e definisce fasi di compilazione che possono trarre vantaggio dalle ottimizzazioni standard e dalle ottimizzazioni specializzate per le reti neurali.", "Questo articolo introduce una DLVM per sfruttare gli aspetti di compilazione di un compilatore tensoriale"]} +{"source": "In this work, we focus on the problem of grounding language by training an agent to follow a set of natural language instructions and navigate to a target object in a 2D grid environment. The agent receives visual information through raw pixels and a natural language instruction telling what task needs to be achieved. Other than these two sources of information, our model does not have any prior information of both the visual and textual modalities and is end-to-end trainable. We develop an attention mechanism for multi-modal fusion of visual and textual modalities that allows the agent to learn to complete the navigation tasks and also achieve language grounding. Our experimental results show that our attention mechanism outperforms the existing multi-modal fusion mechanisms proposed in order to solve the above mentioned navigation task. We demonstrate through the visualization of attention weights that our model learns to correlate attributes of the object referred in the instruction with visual representations and also show that the learnt textual representations are semantically meaningful as they follow vector arithmetic and are also consistent enough to induce translation between instructions in different natural languages. We also show that our model generalizes effectively to unseen scenarios and exhibit zero-shot generalization capabilities. In order to simulate the above described challenges, we introduce a new 2D environment for an agent to jointly learn visual and textual modalities", "target": ["Architettura basata sull'attention per il grounding della lingua attraverso reinforcement learning in un nuovo ambiente a griglia 2D personalizzabile\n", "L'articolo affronta il problema della navigazione data da un'istruzione e propone un approccio per combinare informazioni testuali e visive attraverso un meccanismo di attention", "Questo articolo considera il problema di seguire istruzioni in linguaggio naturale data una visione in prima persona di un ambiente sconosciuto a priori, e propone un metodo con un'architettura neurale.", "Studia il problema della navigazione verso un oggetto target in un ambiente a griglia 2D seguendo una descrizione in linguaggio naturale e ricevendo informazioni visive come pixel grezzi."]} +{"source": "Current end-to-end machine reading and question answering (Q\\&A) models are primarily based on recurrent neural networks (RNNs) with attention. Despite their success, these models are often slow for both training and inference due to the sequential nature of RNNs. We propose a new Q\\&A architecture called QANet, which does not require recurrent networks: Its encoder consists exclusively of convolution and self-attention, where convolution models local interactions and self-attention models global interactions. On the SQuAD dataset, our model is 3x to 13x faster in training and 4x to 9x faster in inference, while achieving equivalent accuracy to recurrent models. The speed-up gain allows us to train the model with much more data. We hence combine our model with data generated by backtranslation from a neural machine translation model. On the SQuAD dataset, our single model, trained with augmented data, achieves 84.6 F1 score on the test set, which is significantly better than the best published F1 score of 81.8.", "target": ["Una semplice architettura composta da convoluzioni e attention raggiunge risultati alla pari con i migliori modelli ricorrenti documentati.", "Un metodo di data augmentation basato sulla parafrasi veloce e ad alte prestazioni e un modello di reading comprehension non ricorrente che usa solo convoluzioni e attention.", "Questo articolo propone di applicare i moduli CNN+self-attention al posto delle LSTM e di migliorare il training del modello RC con le parafrasi dei passaggi generate da un modello neurale di paraphrasing al fine di migliorare le prestazioni RC.", "Questo articolo presenta un modello di reading comprehension utilizzando le convoluzioni e l'attention e propone di aggiungere dati di training attraverso la parafrasi basata sulla traduzione automatica neurale off-the-shelf"]} +{"source": "Convolutional Neural Networks (CNNs) have become the method of choice for learning problems involving 2D planar images. However, a number of problems of recent interest have created a demand for models that can analyze spherical images. Examples include omnidirectional vision for drones, robots, and autonomous cars, molecular regression problems, and global weather and climate modelling. A naive application of convolutional networks to a planar projection of the spherical signal is destined to fail, because the space-varying distortions introduced by such a projection will make translational weight sharing ineffective. In this paper we introduce the building blocks for constructing spherical CNNs. We propose a definition for the spherical cross-correlation that is both expressive and rotation-equivariant. The spherical correlation satisfies a generalized Fourier theorem, which allows us to compute it efficiently using a generalized (non-commutative) Fast Fourier Transform (FFT) algorithm. We demonstrate the computational efficiency, numerical accuracy, and effectiveness of spherical CNNs applied to 3D model recognition and atomization energy regression.", "target": ["Introduciamo Spherical CNN, una rete convoluzionale per segnali sferici, e la applichiamo al riconoscimento di modelli 3D e alla regressione dell'energia molecolare.", "L'articolo propone un framework per la costruzione di reti convoluzionali sferiche basato su una nuova sintesi di diversi concetti esistenti", "Questo articolo si concentra su come estendere le reti neurali convoluzionali per incorporare l'invarianza sferica, e adatta gli strumenti dell'analisi armonica non abeliana per raggiungere questo obiettivo.", "Gli autori sviluppano un nuovo schema per rappresentare i dati sferici da zero"]} +{"source": "We propose a novel method that makes use of deep neural networks and gradient decent to perform automated design on complex real world engineering tasks. Our approach works by training a neural network to mimic the fitness function of a design optimization task and then, using the differential nature of the neural network, perform gradient decent to maximize the fitness. We demonstrate this methods effectiveness by designing an optimized heat sink and both 2D and 3D airfoils that maximize the lift drag ratio under steady state flow conditions. We highlight that our method has two distinct benefits over other automated design approaches. First, evaluating the neural networks prediction of fitness can be orders of magnitude faster then simulating the system of interest. Second, using gradient decent allows the design space to be searched much more efficiently then other gradient free methods. These two strengths work together to overcome some of the current shortcomings of automated design.", "target": ["Un metodo per eseguire la progettazione automatizzata su oggetti del mondo reale come dissipatori di calore e profili alari che fa uso di reti neurali e gradient descent.", "Rete neurale (parametrizzazione e predizione) e gradient descent (backpropagation) per progettare automaticamente i task di ingegneria.", "Questo articolo introduce l'uso di una deep network per approssimare il comportamento di un sistema fisico complesso, e poi progettare dispositivi ottimali ottimizzando questa rete rispetto ai suoi input."]} +{"source": "Methods that align distributions by minimizing an adversarial distance between them have recently achieved impressive results. However, these approaches are difficult to optimize with gradient descent and they often do not converge well without careful hyperparameter tuning and proper initialization. We investigate whether turning the adversarial min-max problem into an optimization problem by replacing the maximization part with its dual improves the quality of the resulting alignment and explore its connections to Maximum Mean Discrepancy. Our empirical results suggest that using the dual formulation for the restricted family of linear discriminators results in a more stable convergence to a desirable solution when compared with the performance of a primal min-max GAN-like objective and an MMD objective under the same restrictions. We test our hypothesis on the problem of aligning two synthetic point clouds on a plane and on a real-image domain adaptation problem on digits. In both cases, the dual formulation yields an iterative procedure that gives more stable and monotonic improvement over time.", "target": ["Proponiamo una versione duale della distanza logistic adversarial per l'allineamento delle feature e mostriamo che produce iterazioni a gradiente più stabili dell'obiettivo min-max.", "L'articolo si occupa di migliorare le GAN a livello computazionale", "Questo articolo studia una formulazione duale di una loss adversarial basata su un upper-bound della loss logistica, e trasforma il problema min max standard dell'adversarial training in un singolo problema di minimizzazione.", "Propone di riformulare l'obiettivo del punto di sella delle GAN (per un discriminatore di regressione logistica) come un problema di minimizzazione dualizzando l'obiettivo di massima verosimiglianza per la regressione logistica regolarizzata"]} +{"source": "There are many applications scenarios for which the computational performance and memory footprint of the prediction phase of Deep Neural Networks (DNNs) need to be optimized. Binary Deep Neural Networks (BDNNs) have been shown to be an effective way of achieving this objective. In this paper, we show how Convolutional Neural Networks (CNNs) can be implemented using binary representations. Espresso is a compact, yet powerful library written in C/CUDA that features all the functionalities required for the forward propagation of CNNs, in a binary file less than 400KB, without any external dependencies. Although it is mainly designed to take advantage of massive GPU parallelism, Espresso also provides an equivalent CPU implementation for CNNs. Espresso provides special convolutional and dense layers for BCNNs, leveraging bit-packing and bit-wise computations for efficient execution. These techniques provide a speed-up of matrix-multiplication routines, and at the same time, reduce memory usage when storing parameters and activations. We experimentally show that Espresso is significantly faster than existing implementations of optimized binary neural networks (~ 2 orders of magnitude). Espresso is released under the Apache 2.0 license and is available at http://github.com/organization/project.", "target": ["Implementazione delle reti neurali binarie allo stato dell'arte in termini di prestazioni computazionali", "L'articolo presenta una libreria scritta in C/CUDA che presenta tutte le funzionalità necessarie per la forward propagation delle BCNN", "Questo paper si basa su Binary-NET e lo espande alle architetture CNN, fornisce ottimizzazioni che migliorano la velocità del forward pass, e fornisce codice ottimizzato per Binary CNN."]} +{"source": "Optimal selection of a subset of items from a given set is a hard problem that requires combinatorial optimization. In this paper, we propose a subset selection algorithm that is trainable with gradient based methods yet achieves near optimal performance via submodular optimization. We focus on the task of identifying a relevant set of sentences for claim verification in the context of the FEVER task. Conventional methods for this task look at sentences on their individual merit and thus do not optimize the informativeness of sentences as a set. We show that our proposed method which builds on the idea of unfolding a greedy algorithm into a computational graph allows both interpretability and gradient based training. The proposed differentiable greedy network (DGN) outperforms discrete optimization algorithms as well as other baseline methods in terms of precision and recall.", "target": ["Proponiamo un algoritmo di subset selection che è addestrabile con metodi basati sul gradiente ma che raggiunge prestazioni quasi ottimali tramite l'ottimizzazione submodulare.", "Propone un modello basato sulla rete neurale che integra la funzione submodulare combinando la tecnica di ottimizzazione basata sul gradiente con un framework submodulare chiamato 'Differentiable Greedy Network' (DGN).", "Propone una rete neurale che mira a selezionare un sottoinsieme di elementi (ad esempio, selezionando k frasi che sono più legate ad un claim da un insieme di documenti recuperati)"]} +{"source": "The joint optimization of representation learning and clustering in the embedding space has experienced a breakthrough in recent years. In spite of the advance, clustering with representation learning has been limited to flat-level categories, which oftentimes involves cohesive clustering with a focus on instance relations. To overcome the limitations of flat clustering, we introduce hierarchically clustered representation learning (HCRL), which simultaneously optimizes representation learning and hierarchical clustering in the embedding space. Specifically, we place a nonparametric Bayesian prior on embeddings to handle dynamic mixture hierarchies under the variational autoencoder framework, and to adopt the generative process of a hierarchical-versioned Gaussian mixture model. Compared with a few prior works focusing on unifying representation learning and hierarchical clustering, HCRL is the first model to consider a generation of deep embeddings from every component of the hierarchy, not just leaf components. This generation process enables more meaningful separations and mergers of clusters via branches in a hierarchy. In addition to obtaining hierarchically clustered embeddings, we can reconstruct data by the various abstraction levels, infer the intrinsic hierarchical structure, and learn the level-proportion features. We conducted evaluations with image and text domains, and our quantitative analyses showed competent likelihoods and the best accuracies compared with the baselines.", "target": ["Introduciamo lo hierarchically clustered representation learning (HCRL), che ottimizza simultaneamente il representation learning e il clustering gerarchico nello spazio di embedding.", "L'articolo propone di usare il CRP annidato come un modello di clustering piuttosto che come un topic model", "Presenta un nuovo metodo di clustering gerarchico su uno spazio di embedding in cui sia lo spazio di embedding che il clustering gerarchico sono appresi simultaneamente"]} +{"source": "We introduce a novel geometric perspective and unsupervised model augmentation framework for transforming traditional deep (convolutional) neural networks into adversarially robust classifiers. Class-conditional probability densities based on Bayesian nonparametric mixtures of factor analyzers (BNP-MFA) over the input space are used to design soft decision labels for feature to label isometry. Classconditional distributions over features are also learned using BNP-MFA to develop plug-in maximum a posterior (MAP) classifiers to replace the traditional multinomial logistic softmax classification layers. This novel unsupervised augmented framework, which we call geometrically robust networks (GRN), is applied to CIFAR-10, CIFAR-100, and to Radio-ML (a time series dataset for radio modulation recognition). We demonstrate the robustness of GRN models to adversarial attacks from fast gradient sign method, Carlini-Wagner, and projected gradient descent.", "target": ["Sviluppiamo un framework di augmentation dell'unsupervised learning statistico-geometrico per deep neural network per renderle robuste agli adversarial attack.", "Trasforma le deep neural network tradizionali in classificatori robusti adversarial utilizzando le GRN", "Propone una difesa basata su distribuzioni di feature class-conditional per trasformare le deep neural network in classificatori robusti"]} +{"source": "Reinforcement learning in environments with large state-action spaces is challenging, as exploration can be highly inefficient. Even if the dynamics are simple, the optimal policy can be combinatorially hard to discover. In this work, we propose a hierarchical approach to structured exploration to improve the sample efficiency of on-policy exploration in large state-action spaces. The key idea is to model a stochastic policy as a hierarchical latent variable model, which can learn low-dimensional structure in the state-action space, and to define exploration by sampling from the low-dimensional latent space. This approach enables lower sample complexity, while preserving policy expressivity. In order to make learning tractable, we derive a joint learning and exploration strategy by combining hierarchical variational inference with actor-critic learning. The benefits of our learning approach are that 1) it is principled, 2) simple to implement, 3) easily scalable to settings with many actions and 4) easily composable with existing deep learning approaches. We demonstrate the effectiveness of our approach on learning a deep centralized multi-agent policy, as multi-agent environments naturally have an exponentially large state-action space. In this setting, the latent hierarchy implements a form of multi-agent coordination during exploration and execution (MACE). We demonstrate empirically that MACE can more efficiently learn optimal policies in challenging multi-agent games with a large number (~20) of agents, compared to conventional baselines. Moreover, we show that our hierarchical structure leads to meaningful agent coordination.", "target": ["Rendere più efficiente il deep reinforcement learning in grandi spazi di stato-azione usando l'esplorazione strutturata con deep hierarchical policy.", "Un metodo per coordinare il comportamento degli agenti usando policy che hanno una struttura latente condivisa, un metodo di ottimizzazione variazionale delle policy per ottimizzare le policy coordinate, e una derivazione dell'update variazionale e gerarchico degli autori.", "Questo articolo suggerisce un'innovazione algoritmica costituita da variabili latenti gerarchiche per l'esplorazione coordinata in ambienti multi-agente"]} +{"source": "Much attention has been devoted recently to the generalization puzzle in deep learning: large, deep networks can generalize well, but existing theories bounding generalization error are exceedingly loose, and thus cannot explain this striking performance. Furthermore, a major hope is that knowledge may transfer across tasks, so that multi-task learning can improve generalization on individual tasks. However we lack analytic theories that can quantitatively predict how the degree of knowledge transfer depends on the relationship between the tasks. We develop an analytic theory of the nonlinear dynamics of generalization in deep linear networks, both within and across tasks. In particular, our theory provides analytic solutions to the training and testing error of deep networks as a function of training time, number of examples, network size and initialization, and the task structure and SNR. Our theory reveals that deep networks progressively learn the most important task structure first, so that generalization error at the early stopping time primarily depends on task structure and is independent of network size. This suggests any tight bound on generalization error must take into account task structure, and explains observations about real data being learned faster than random data. Intriguingly our theory also reveals the existence of a learning algorithm that proveably out-performs neural network training through gradient descent. Finally, for transfer learning, our theory reveals that knowledge transfer depends sensitively, but computably, on the SNRs and input feature alignments of pairs of tasks.", "target": ["Forniamo molte intuizioni sulla generalizzazione delle reti neurali dal caso lineare trattabile dalla teoria.", "Gli autori studiano un semplice modello di reti lineari per comprendere la generalizzazione e il transfer learning"]} +{"source": "We conduct a mathematical analysis on the Batch normalization (BN) effect on gradient backpropagation in residual network training in this work, which is believed to play a critical role in addressing the gradient vanishing/explosion problem. Specifically, by analyzing the mean and variance behavior of the input and the gradient in the forward and backward passes through the BN and residual branches, respectively, we show that they work together to confine the gradient variance to a certain range across residual blocks in backpropagation. As a result, the gradient vanishing/explosion problem is avoided. Furthermore, we use the same analysis to discuss the tradeoff between depth and width of a residual network and demonstrate that shallower yet wider resnets have stronger learning performance than deeper yet thinner resnets.", "target": ["La batch normalization mantiene la varianza del gradiente durante l'allenamento, stabilizzando così l'ottimizzazione.", "Questo articolo ha analizzato l'effetto della batch normalization sulla backpropagation del gradiente nelle reti residuali"]} +{"source": "To study how mental object representations are related to behavior, we estimated sparse, non-negative representations of objects using human behavioral judgments on images representative of 1,854 object categories. These representations predicted a latent similarity structure between objects, which captured most of the explainable variance in human behavioral judgments. Individual dimensions in the low-dimensional embedding were found to be highly reproducible and interpretable as conveying degrees of taxonomic membership, functionality, and perceptual attributes. We further demonstrated the predictive power of the embeddings for explaining other forms of human behavior, including categorization, typicality judgments, and feature ratings, suggesting that the dimensions reflect human conceptual representations of objects beyond the specific task.", "target": ["I giudizi comportamentali umani sono utilizzati per ottenere rappresentazioni sparse e interpretabili degli oggetti che si generalizzano ad altri task", "Questo articolo descrive un esperimento su larga scala sulle rappresentazioni umane semantiche o di oggetti e un modello di tali rappresentazioni.", "Questo articolo sviluppa un nuovo sistema di rappresentazione per le rappresentazioni di oggetti dal training sui dati raccolti da giudizi umani di immagini di tipo odd-one-out.", "Un nuovo approccio per apprendere uno spazio semantico sparso, positivo e interpretabile che massimizza i giudizi umani di somiglianza attraverso il training per massimizzare specificamente la predizione dei giudizi umani di somiglianza."]} +{"source": "We frame Question Answering (QA) as a Reinforcement Learning task, an approach that we call Active Question Answering. We propose an agent that sits between the user and a black box QA system and learns to reformulate questions to elicit the best possible answers. The agent probes the system with, potentially many, natural language reformulations of an initial question and aggregates the returned evidence to yield the best answer. The reformulation system is trained end-to-end to maximize answer quality using policy gradient. We evaluate on SearchQA, a dataset of complex questions extracted from Jeopardy!. The agent outperforms a state-of-the-art base model, playing the role of the environment, and other benchmarks. We also analyze the language that the agent has learned while interacting with the question answering system. We find that successful question reformulations look quite different from natural language paraphrases. The agent is able to discover non-trivial reformulation strategies that resemble classic information retrieval techniques such as term re-weighting (tf-idf) and stemming.", "target": ["Proponiamo un agente che si pone tra l'utente e un sistema di question answering a scatola nera e che impara a riformulare le domande per ottenere le migliori risposte possibili", "Questo articolo propone un active question answering attraverso un approccio di reinforcement learning che impara a riformulare le domande in modo da fornire le migliori risposte possibili.", "Descrive chiaramente come i ricercatori hanno progettato e addestrato attivamente due modelli per la riformulazione delle domande e la selezione delle risposte durante gli episodi di question answering"]} +{"source": "Most deep latent factor models choose simple priors for simplicity, tractability or not knowing what prior to use. Recent studies show that the choice of the prior may have a profound effect on the expressiveness of the model, especially when its generative network has limited capacity. In this paper, we propose to learn a proper prior from data for adversarial autoencoders (AAEs). We introduce the notion of code generators to transform manually selected simple priors into ones that can better characterize the data distribution. Experimental results show that the proposed model can generate better image quality and learn better disentangled representations than AAEs in both supervised and unsupervised settings. Lastly, we present its ability to do cross-domain translation in a text-to-image synthesis task.", "target": ["Apprendimento di prior per adversarial autoencoder", "Propone una semplice estensione degli adversarial autoencoder per la generazione condizionale di immagini.", "Si concentra sugli adversarial autoencoder e introduce una rete di generatori di codice per trasformare un semplice prior in uno che insieme al generatore può adattarsi meglio alla distribuzione dei dati"]} +{"source": "In the past few years, various advancements have been made in generative models owing to the formulation of Generative Adversarial Networks (GANs). GANs have been shown to perform exceedingly well on a wide variety of tasks pertaining to image generation and style transfer. In the field of Natural Language Processing, word embeddings such as word2vec and GLoVe are state-of-the-art methods for applying neural network models on textual data. Attempts have been made for utilizing GANs with word embeddings for text generation. This work presents an approach to text generation using Skip-Thought sentence embeddings in conjunction with GANs based on gradient penalty functions and f-measures. The results of using sentence embeddings with GANs for generating text conditioned on input information are comparable to the approaches where word embeddings are used.", "target": ["Generazione di testo usando gli embedding di frasi dai vettori skip-thought con l'aiuto delle generative adversarial network.", "Descrive l'applicazione delle generative adversarial network per la modellazione di dati testuali con l'aiuto di vettori skip-thought e gli esperimenti con diversi tipi di GAN per due diversi dataset."]} +{"source": "The novel \\emph{Unbiased Online Recurrent Optimization} (UORO) algorithm allows for online learning of general recurrent computational graphs such as recurrent network models. It works in a streaming fashion and avoids backtracking through past activations and inputs. UORO is computationally as costly as \\emph{Truncated Backpropagation Through Time} (truncated BPTT), a widespread algorithm for online learning of recurrent networks \\cite{jaeger2002tutorial}. UORO is a modification of \\emph{NoBackTrack} \\cite{DBLP:journals/corr/OllivierC15} that bypasses the need for model sparsity and makes implementation easy in current deep learning frameworks, even for complex models. Like NoBackTrack, UORO provides unbiased gradient estimates; unbiasedness is the core hypothesis in stochastic gradient descent theory, without which convergence to a local optimum is not guaranteed. On the contrary, truncated BPTT does not provide this property, leading to possible divergence. On synthetic tasks where truncated BPTT is shown to diverge, UORO converges. For instance, when a parameter has a positive short-term but negative long-term influence, truncated BPTT diverges unless the truncation span is very significantly longer than the intrinsic temporal range of the interactions, while UORO performs well thanks to the unbiasedness of its gradients.", "target": ["Introduce una stima del gradiente online, unbiased e facilmente implementabile per modelli ricorrenti.", "Gli autori introducono un nuovo approccio all'apprendimento online dei parametri delle reti neurali ricorrenti da lunghe sequenze che supera la limitazione della backpropagation troncata nel tempo", "Questo articolo affronta il training online di RNN, e propone una modifica a RTRL e di usare l'approccio forward per il calcolo del gradiente."]} +{"source": "We present a deep learning-based method for super-resolving coarse (low-resolution) labels assigned to groups of image pixels into pixel-level (high-resolution) labels, given the joint distribution between those low- and high-resolution labels. This method involves a novel loss function that minimizes the distance between a distribution determined by a set of model outputs and the corresponding distribution given by low-resolution labels over the same set of outputs. This setup does not require that the high-resolution classes match the low-resolution classes and can be used in high-resolution semantic segmentation tasks where high-resolution labeled data is not available. Furthermore, our proposed method is able to utilize both data with low-resolution labels and any available high-resolution labels, which we show improves performance compared to a network trained only with the same amount of high-resolution data. We test our proposed algorithm in a challenging land cover mapping task to super-resolve labels at a 30m resolution to a separate set of labels at a 1m resolution. We compare our algorithm with models that are trained on high-resolution data and show that 1) we can achieve similar performance using only low-resolution data; and 2) we can achieve better performance when we incorporate a small amount of high-resolution data in our training. We also test our approach on a medical imaging problem, resolving low-resolution probability maps into high-resolution segmentation of lymphocytes with accuracy equal to that of fully supervised models.", "target": ["Super-risoluzione di label coarse in label a livello di pixel, applicate a immagini aeree e scansioni mediche.", "Un metodo per risolvere le label di segmentazione coarse a bassa risoluzione se la distribuzione congiunta delle label a bassa e alta risoluzione è nota."]} +{"source": "We propose a novel framework for combining datasets via alignment of their associated intrinsic dimensions. Our approach assumes that the two datasets are sampled from a common latent space, i.e., they measure equivalent systems. Thus, we expect there to exist a natural (albeit unknown) alignment of the data manifolds associated with the intrinsic geometry of these datasets, which are perturbed by measurement artifacts in the sampling process. Importantly, we do not assume any individual correspondence (partial or complete) between data points. Instead, we rely on our assumption that a subset of data features have correspondence across datasets. We leverage this assumption to estimate relations between intrinsic manifold dimensions, which are given by diffusion map coordinates over each of the datasets. We compute a correlation matrix between diffusion coordinates of the datasets by considering graph (or manifold) Fourier coefficients of corresponding data features. We then orthogonalize this correlation matrix to form an isometric transformation between the diffusion maps of the datasets. Finally, we apply this transformation to the diffusion coordinates and construct a unified diffusion geometry of the datasets together. We show that this approach successfully corrects misalignment artifacts, and allows for integrated data.", "target": ["Proponiamo un metodo per allineare le feature latenti apprese da diversi dataset utilizzando le correlazioni armoniche.", "Propone l'uso di corrispondenze di feature per preformare l'allineamento manifold tra batch di dati dagli stessi sample per evitare la raccolta di misure rumorose."]} +{"source": "Reinforcement learning (RL) has proven to be a powerful paradigm for deriving complex behaviors from simple reward signals in a wide range of environments. When applying RL to continuous control agents in simulated physics environments, the body is usually considered to be part of the environment. However, during evolution the physical body of biological organisms and their controlling brains are co-evolved, thus exploring a much larger space of actuator/controller configurations. Put differently, the intelligence does not reside only in the agent's mind, but also in the design of their body. We propose a method for uncovering strong agents, consisting of a good combination of a body and policy, based on combining RL with an evolutionary procedure. Given the resulting agent, we also propose an approach for identifying the body changes that contributed the most to the agent performance. We use the Shapley value from cooperative game theory to find the fair contribution of individual components, taking into account synergies between components. We evaluate our methods in an environment similar to the the recently proposed Robo-Sumo task, where agents in a 3D environment with simulated physics compete in tipping over their opponent or pushing them out of the arena. Our results show that the proposed methods are indeed capable of generating strong agents, significantly outperforming baselines that focus on optimizing the agent policy alone. A video is available at: www.youtube.com/watch?v=eei6Rgom3YY", "target": ["L'evoluzione della forma del corpo negli agenti controllati da RL migliora le loro prestazioni (e aiuta l'apprendimento)", "Algoritmo PEOM che incorpora il valore Shapley per accelerare l'evoluzione identificando il contributo di ogni parte del corpo"]} +{"source": "Many practical reinforcement learning problems contain catastrophic states that the optimal policy visits infrequently or never. Even on toy problems, deep reinforcement learners periodically revisit these states, once they are forgotten under a new policy. In this paper, we introduce intrinsic fear, a learned reward shaping that accelerates deep reinforcement learning and guards oscillating policies against periodic catastrophes. Our approach incorporates a second model trained via supervised learning to predict the probability of imminent catastrophe. This score acts as a penalty on the Q-learning objective. Our theoretical analysis demonstrates that the perturbed objective yields the same average return under strong assumptions and an $\\epsilon$-close average return under weaker assumptions. Our analysis also shows robustness to classification errors. Equipped with intrinsic fear, our DQNs solve the toy environments and improve on the Atari games Seaquest, Asteroids, and Freeway.", "target": ["Modellare la reward con la motivazione intrinseca per evitare stati catastrofici e mitigare il catastrophic forgetting.", "Un algoritmo di RL che combina l'algoritmo DQN con un fear model addestrato in parallelo per prevedere gli stati catastrofici.", "L'articolo studia il catastrophic forgetting nel RL, enfatizzando i task in cui una DQN è in grado di imparare ad evitare eventi catastrofici a patto che eviti il forgetting."]} +{"source": "Convolution is an efficient technique to obtain abstract feature representations using hierarchical layers in deep networks. Although performing convolution in Euclidean geometries is fairly straightforward, its extension to other topological spaces---such as a sphere S^2 or a unit ball B^3---entails unique challenges. In this work, we propose a novel `\"volumetric convolution\" operation that can effectively convolve arbitrary functions in B^3. We develop a theoretical framework for \"volumetric convolution\" based on Zernike polynomials and efficiently implement it as a differentiable and an easily pluggable layer for deep networks. Furthermore, our formulation leads to derivation of a novel formula to measure the symmetry of a function in B^3 around an arbitrary axis, that is useful in 3D shape analysis tasks. We demonstrate the efficacy of proposed volumetric convolution operation on a possible use-case i.e., 3D object recognition task.", "target": ["Un nuovo operatore di convoluzione per l'apprendimento automatico della rappresentazione all'interno della sfera unitaria", "Questo lavoro è legato ai recenti articoli sulle reti sferiche CNN e SE(n) equivarianti ed estende le idee precedenti ai dati volumetrici nella sfera unitaria.", "Propone l'uso di convoluzioni volumetriche su reti di convoluzioni per imparare la sfera unitaria e discute la metodologia e i risultati del processo."]} +{"source": "Learning in environments with large state and action spaces, and sparse rewards, can hinder a Reinforcement Learning (RL) agent’s learning through trial-and-error. For instance, following natural language instructions on the Web (such as booking a flight ticket) leads to RL settings where input vocabulary and number of actionable elements on a page can grow very large. Even though recent approaches improve the success rate on relatively simple environments with the help of human demonstrations to guide the exploration, they still fail in environments where the set of possible instructions can reach millions. We approach the aforementioned problems from a different perspective and propose guided RL approaches that can generate unbounded amount of experience for an agent to learn from. Instead of learning from a complicated instruction with a large vocabulary, we decompose it into multiple sub-instructions and schedule a curriculum in which an agent is tasked with a gradually increasing subset of these relatively easier sub-instructions. In addition, when the expert demonstrations are not available, we propose a novel meta-learning framework that generates new instruction following tasks and trains the agent more effectively. We train DQN, deep reinforcement learning agent, with Q-value function approximated with a novel QWeb neural network architecture on these smaller, synthetic instructions. We evaluate the ability of our agent to generalize to new instructions onWorld of Bits benchmark, on forms with up to 100 elements, supporting 14 million possible instructions. The QWeb agent outperforms the baseline without using any human demonstration achieving 100% success rate on several difficult environments.", "target": ["Addestriamo policy di reinforcement learning utilizzando l'aumento della reward, il curriculum learning e il meta-learning per navigare con successo le pagine web.", "Sviluppa un metodo di curriculum learning per addestrare un agente di RL a navigare nel web, basato sull'idea di decomporre un'istruzione in più sotto-istruzioni."]} +{"source": "Labeled text classification datasets are typically only available in a few select languages. In order to train a model for e.g news categorization in a language $L_t$ without a suitable text classification dataset there are two options. The first option is to create a new labeled dataset by hand, and the second option is to transfer label information from an existing labeled dataset in a source language $L_s$ to the target language $L_t$. In this paper we propose a method for sharing label information across languages by means of a language independent text encoder. The encoder will give almost identical representations to multilingual versions of the same text. This means that labeled data in one language can be used to train a classifier that works for the rest of the languages. The encoder is trained independently of any concrete classification task and can therefore subsequently be used for any classification task. We show that it is possible to obtain good performance even in the case where only a comparable corpus of texts is available.", "target": ["Classificazione del testo cross language mediante codifica universale", "Questo articolo propone un approccio alla classificazione cross-linguale dei testi attraverso l'uso di corpora comparabili.", "Apprendimento di cross-lingual embedding e training di un classificatore utilizzando dati annotati nella lingua di origine per affrontare l'apprendimento di un cross-language text categorizer senza informazioni annotate nella lingua di destinazione"]} +{"source": "Syntax is a powerful abstraction for language understanding. Many downstream tasks require segmenting input text into meaningful constituent chunks (e.g., noun phrases or entities); more generally, models for learning semantic representations of text benefit from integrating syntax in the form of parse trees (e.g., tree-LSTMs). Supervised parsers have traditionally been used to obtain these trees, but lately interest has increased in unsupervised methods that induce syntactic representations directly from unlabeled text. To this end, we propose the deep inside-outside recursive autoencoder (DIORA), a fully-unsupervised method for discovering syntax that simultaneously learns representations for constituents within the induced tree. Unlike many prior approaches, DIORA does not rely on supervision from auxiliary downstream tasks and is thus not constrained to particular domains. Furthermore, competing approaches do not learn explicit phrase representations along with tree structures, which limits their applicability to phrase-based tasks. Extensive experiments on unsupervised parsing, segmentation, and phrase clustering demonstrate the efficacy of our method. DIORA achieves the state of the art in unsupervised parsing (46.9 F1) on the benchmark WSJ dataset.", "target": ["In questo lavoro proponiamo deep inside-outside recursive auto-encoder (DIORA) un metodo completamente unsupervised per scoprire la sintassi e contemporaneamente imparare rappresentazioni per i costituenti scoperti.", "Un modello neurale ad albero latente addestrato con un obiettivo di autoencoding che raggiunge lo stato dell'arte nell'unsupervised constituency parsing e cattura la struttura sintattica meglio di altri modelli ad albero latente.", "L'articolo propone un modello per unsupervised dependency parsing (induzione dell'albero latente) che è basato su una combinazione dell'algoritmo inside-outside con la modellazione neurale (autoencoder ricorsivi)."]} +{"source": "Careful tuning of the learning rate, or even schedules thereof, can be crucial to effective neural net training. There has been much recent interest in gradient-based meta-optimization, where one tunes hyperparameters, or even learns an optimizer, in order to minimize the expected loss when the training procedure is unrolled. But because the training procedure must be unrolled thousands of times, the meta-objective must be defined with an orders-of-magnitude shorter time horizon than is typical for neural net training. We show that such short-horizon meta-objectives cause a serious bias towards small step sizes, an effect we term short-horizon bias. We introduce a toy problem, a noisy quadratic cost function, on which we analyze short-horizon bias by deriving and comparing the optimal schedules for short and long time horizons. We then run meta-optimization experiments (both offline and online) on standard benchmark datasets, showing that meta-optimization chooses too small a learning rate by multiple orders of magnitude, even when run with a moderately long time horizon (100 steps) typical of work in the area. We believe short-horizon bias is a fundamental problem that needs to be addressed if meta-optimization is to scale to practical neural net training regimes.", "target": ["Indaghiamo il bias nell'obiettivo di meta-ottimizzazione short-horizon.", "Questo articolo propone un modello e un problema semplificato per dimostrare lo short-horizon bias della meta-ottimizzazione del learning rate.", "Questo articolo studia la questione della truncated backpropagation per la meta-ottimizzazione attraverso una serie di esperimenti su un problema giocattolo"]} +{"source": "Mainstream captioning models often follow a sequential structure to generate cap- tions, leading to issues such as introduction of irrelevant semantics, lack of diversity in the generated captions, and inadequate generalization performance. In this paper, we present an alternative paradigm for image captioning, which factorizes the captioning procedure into two stages: (1) extracting an explicit semantic represen- tation from the given image; and (2) constructing the caption based on a recursive compositional procedure in a bottom-up manner. Compared to conventional ones, our paradigm better preserves the semantic content through an explicit factorization of semantics and syntax. By using the compositional generation procedure, caption construction follows a recursive structure, which naturally fits the properties of human language. Moreover, the proposed compositional procedure requires less data to train, generalizes better, and yields more diverse captions.", "target": ["Un modo gerarchico e compositivo di generare didascalie", "Questo articolo presenta un metodo più interpretabile per l'image captioning."]} +{"source": "While many approaches to make neural networks more fathomable have been proposed, they are restricted to interrogating the network with input data. Measures for characterizing and monitoring structural properties, however, have not been developed. In this work, we propose neural persistence, a complexity measure for neural network architectures based on topological data analysis on weighted stratified graphs. To demonstrate the usefulness of our approach, we show that neural persistence reflects best practices developed in the deep learning community such as dropout and batch normalization. Moreover, we derive a neural persistence-based stopping criterion that shortens the training process while achieving comparable accuracies as early stopping based on validation loss.", "target": ["Sviluppiamo una nuova misura di complessità topologica per le deep neural network e dimostriamo che cattura le loro proprietà salienti.", "Questo articolo propone la nozione di persistenza neurale, una misura topologica per assegnare punteggi ai layer completamente connessi in una rete neurale.", "L'articolo propone di analizzare la complessità di una rete neurale usando la sua omologia persistente di ordine zero."]} +{"source": "Deep neural networks (DNNs) are vulnerable to adversarial examples, which are carefully crafted instances aiming to cause prediction errors for DNNs. Recent research on adversarial examples has examined local neighborhoods in the input space of DNN models. However, previous work has limited what regions to consider, focusing either on low-dimensional subspaces or small balls. In this paper, we argue that information from larger neighborhoods, such as from more directions and from greater distances, will better characterize the relationship between adversarial examples and the DNN models. First, we introduce an attack, OPTMARGIN, which generates adversarial examples robust to small perturbations. These examples successfully evade a defense that only considers a small ball around an input instance. Second, we analyze a larger neighborhood around input instances by looking at properties of surrounding decision boundaries, namely the distances to the boundaries and the adjacent classes. We find that the boundaries around these adversarial examples do not resemble the boundaries around benign examples. Finally, we show that, under scrutiny of the surrounding decision boundaries, our OPTMARGIN examples do not convincingly mimic benign examples. Although our experiments are limited to a few specific attacks, we hope these findings will motivate new, more evasive attacks and ultimately, effective defenses.", "target": ["Guardare i confini decisionali intorno a un input ti dà più informazioni di un piccolo neighborhood fisso", "Gli autori presentano un nuovo attacco per la generazione di adversarial example in cui attaccano i classificatori creati classificando casualmente piccole perturbazioni L2", "Un nuovo approccio per generare adversarial attack a una rete neurale e un metodo per difendere una rete neurale da questi attacchi."]} +{"source": "Machine learning models are usually tuned by nesting optimization of model weights inside the optimization of hyperparameters. We give a method to collapse this nested optimization into joint stochastic optimization of both weights and hyperparameters. Our method trains a neural network to output approximately optimal weights as a function of hyperparameters. We show that our method converges to locally optimal weights and hyperparameters for sufficiently large hypernets. We compare this method to standard hyperparameter optimization strategies and demonstrate its effectiveness for tuning thousands of hyperparameters.", "target": ["Addestriamo una rete neurale per produrre pesi approssimativamente ottimali in funzione degli iperparametri.", "Iper-reti per l'ottimizzazione degli iper-parametri nelle reti neurali."]} +{"source": "Estimating covariances between financial assets plays an important role in risk management. In practice, when the sample size is small compared to the number of variables, the empirical estimate is known to be very unstable. Here, we propose a novel covariance estimator based on the Gaussian Process Latent Variable Model (GP-LVM). Our estimator can be considered as a non-linear extension of standard factor models with readily interpretable parameters reminiscent of market betas. Furthermore, our Bayesian treatment naturally shrinks the sample covariance matrix towards a more structured matrix given by the prior and thereby systematically reduces estimation errors. Finally, we discuss some financial applications of the GP-LVM model.", "target": ["Stima della matrice di covarianza delle attività finanziarie con modelli a variabili latenti con processo gaussiano", "Illustra come il Gaussian Process Latent Variable Model (GP-LVM) può sostituire i classici modelli a fattori lineari per la stima delle matrici di covarianza nei problemi di ottimizzazione del portafoglio.", "Questo articolo utilizza GPLVM standard per modellare la struttura di covarianza e una rappresentazione dello spazio latente delle serie temporali finanziarie S&P500, per ottimizzare i portafogli e prevedere i valori mancanti.", "Questo articolo propone di utilizzare una GPLVM per modellare i rendimenti finanziari"]} +{"source": "We study how, in generative adversarial networks, variance in the discriminator's output affects the generator's ability to learn the data distribution. In particular, we contrast the results from various well-known techniques for training GANs when the discriminator is near-optimal and updated multiple times per update to the generator. As an alternative, we propose an additional method to train GANs by explicitly modeling the discriminator's output as a bi-modal Gaussian distribution over the real/fake indicator variables. In order to do this, we train the Gaussian classifier to match the target bi-modal distribution implicitly through meta-adversarial training. We observe that our new method, when trained together with a strong discriminator, provides meaningful, non-vanishing gradients.", "target": ["Introduciamo l'apprendimento meta-adversarial, una nuova tecnica per regolarizzare le GAN, e proponiamo un metodo di training controllando esplicitamente la distribuzione dell'output del discriminatore.", "L'articolo propone l'adversarial learning che regolarizza la varianza per il training delle GAN per assicurare che il gradiente del generatore non svanisca"]} +{"source": "We introduce NoisyNet, a deep reinforcement learning agent with parametric noise added to its weights, and show that the induced stochasticity of the agent’s policy can be used to aid efficient exploration. The parameters of the noise are learned with gradient descent along with the remaining network weights. NoisyNet is straightforward to implement and adds little computational overhead. We find that replacing the conventional exploration heuristics for A3C, DQN and Dueling agents (entropy reward and epsilon-greedy respectively) with NoisyNet yields substantially higher scores for a wide range of Atari games, in some cases advancing the agent from sub to super-human performance.", "target": ["Un agente di deep reinforcement learning con rumore parametrico aggiunto ai suoi pesi può essere utilizzato per aiutare l'esplorazione efficiente.", "Questo articolo introduce le NoisyNet, reti neurali i cui parametri sono perturbati da una funzione di rumore parametrico, che ottengono un sostanziale miglioramento delle prestazioni rispetto agli algoritmi base di deep reinforcement learning.", "Nuovo metodo di esplorazione per deep RL iniettando rumore nei pesi delle deep network, con il rumore che assume varie forme"]} +{"source": "Localization is the problem of estimating the location of an autonomous agent from an observation and a map of the environment. Traditional methods of localization, which filter the belief based on the observations, are sub-optimal in the number of steps required, as they do not decide the actions taken by the agent. We propose \"Active Neural Localizer\", a fully differentiable neural network that learns to localize efficiently. The proposed model incorporates ideas of traditional filtering-based localization methods, by using a structured belief of the state with multiplicative interactions to propagate belief, and combines it with a policy model to minimize the number of steps required for localization. Active Neural Localizer is trained end-to-end with reinforcement learning. We use a variety of simulation environments for our experiments which include random 2D mazes, random mazes in the Doom game engine and a photo-realistic environment in the Unreal game engine. The results on the 2D environments show the effectiveness of the learned policy in an idealistic setting while results on the 3D environments demonstrate the model's capability of learning the policy and perceptual model jointly from raw-pixel based RGB observations. We also show that a model trained on random textures in the Doom environment generalizes well to a photo-realistic office space environment in the Unreal engine.", "target": ["\"Active Neural Localizer\", una rete neurale completamente differenziabile che impara a localizzare in modo efficiente utilizzando deep reinforcement learning.", "Questo articolo formula il problema della localizzazione su una mappa nota usando una belief network come un problema di RL in cui l'obiettivo dell'agente è di minimizzare il numero di step per localizzarsi.", "Questo è un paper chiaro e interessante che costruisce una rete parametrizzata per selezionare azioni per un robot in un ambiente simulato"]} +{"source": "Machine translation is an important real-world application, and neural network-based AutoRegressive Translation (ART) models have achieved very promising accuracy. Due to the unparallelizable nature of the autoregressive factorization, ART models have to generate tokens one by one during decoding and thus suffer from high inference latency. Recently, Non-AutoRegressive Translation (NART) models were proposed to reduce the inference time. However, they could only achieve inferior accuracy compared with ART models. To improve the accuracy of NART models, in this paper, we propose to leverage the hints from a well-trained ART model to train the NART model. We define two hints for the machine translation task: hints from hidden states and hints from word alignments, and use such hints to regularize the optimization of NART models. Experimental results show that the NART model trained with hints could achieve significantly better translation performance than previous NART models on several tasks. In particular, for the WMT14 En-De and De-En task, we obtain BLEU scores of 25.20 and 29.52 respectively, which largely outperforms the previous non-autoregressive baselines. It is even comparable to a strong LSTM-based ART model (24.60 on WMT14 En-De), but one order of magnitude faster in inference.", "target": ["Sviluppiamo un algoritmo di training per modelli di traduzione automatica non autoregressivi, ottenendo un'accuratezza paragonabile alle baseline autoregressive forti, essendo però un ordine di grandezza più veloce nell'inferenza.", "Distilla la conoscenza dagli hidden state intermedi e dai pesi di attention per migliorare la traduzione automatica neurale non autoregressiva.", "Propone di sfruttare il modello autoregressivo ben addestrato per informare gli hidden state e l'allineamento delle parole dei modelli di traduzione automatica non autoregressivi."]} +{"source": "Artificial neural networks are built on the basic operation of linear combination and non-linear activation function. Theoretically this structure can approximate any continuous function with three layer architecture. But in practice learning the parameters of such network can be hard. Also the choice of activation function can greatly impact the performance of the network. In this paper we are proposing to replace the basic linear combination operation with non-linear operations that do away with the need of additional non-linear activation function. To this end we are proposing the use of elementary morphological operations (dilation and erosion) as the basic operation in neurons. We show that these networks (Denoted as Morph-Net) with morphological operations can approximate any smooth function requiring less number of parameters than what is necessary for normal neural networks. The results show that our network perform favorably when compared with similar structured network. We have carried out our experiments on MNIST, Fashion-MNIST, CIFAR10 and CIFAR100.", "target": ["Usando le operazioni morfologiche (dilatazione ed erosione) abbiamo definito una classe di reti che può approssimare qualsiasi funzione continua.", "Questo articolo propone di sostituire le unità RELU/tanh standard con una combinazione di operazioni di dilatazione ed erosione, osservando che il nuovo operatore crea più iperpiani e ha più potenza espressiva.", "Gli autori introducono Morph-Net, una rete neurale a singolo layer in cui la mappatura viene eseguita utilizzando la dilatazione e l'erosione morfologica."]} +{"source": "With the rapidly scaling up of deep neural networks (DNNs), extensive research studies on network model compression such as weight pruning have been performed for efficient deployment. This work aims to advance the compression beyond the weights to the activations of DNNs. We propose the Integral Pruning (IP) technique which integrates the activation pruning with the weight pruning. Through the learning on the different importance of neuron responses and connections, the generated network, namely IPnet, balances the sparsity between activations and weights and therefore further improves execution efficiency. The feasibility and effectiveness of IPnet are thoroughly evaluated through various network models with different activation functions and on different datasets. With <0.5% disturbance on the testing accuracy, IPnet saves 71.1% ~ 96.35% of computation cost, compared to the original dense models with up to 5.8x and 10x reductions in activation and weight numbers, respectively.", "target": ["Questo lavoro fa avanzare la compressione DNN oltre i pesi applicandola alle attivazioni integrando il pruning delle attivazioni con il pruning dei pesi.", "Un metodo di compressione integrale del modello che gestisce sia il pruning dei pesi che delle attivazioni, portando ad una computazione delle reti più efficiente e a una riduzione efficace del numero di moltiplicazioni e accumulazioni.", "Questo articolo presenta un nuovo approccio per ridurre il costo di calcolo delle deep neural network integrando il pruning delle attivazioni a quello dei pesi e mostra che le tecniche comuni di pruning applicate solo ai pesi aumentano il numero di attivazioni non nulle dopo ReLU."]} +{"source": "The Variational Auto Encoder (VAE) is a popular generative latent variable model that is often applied for representation learning. Standard VAEs assume continuous valued latent variables and are trained by maximization of the evidence lower bound (ELBO). Conventional methods obtain a differentiable estimate of the ELBO with reparametrized sampling and optimize it with Stochastic Gradient Descend (SGD). However, this is not possible if we want to train VAEs with discrete valued latent variables, since reparametrized sampling is not possible. Till now, there exist no simple solutions to circumvent this problem. In this paper, we propose an easy method to train VAEs with binary or categorically valued latent representations. Therefore, we use a differentiable estimator for the ELBO which is based on importance sampling. In experiments, we verify the approach and train two different VAEs architectures with Bernoulli and Categorically distributed latent representations on two different benchmark datasets.", "target": ["Proponiamo un metodo semplice per addestrare Variational Auto Encoder (VAE) con rappresentazioni latenti discrete, utilizzando l'importance sampling", "Introdurre una distribuzione di importance sampling e usare sample dalla distribuzione per calcolare la stima pesata d'importanza del gradiente", "Questo articolo propone di usare l'importance sampling per ottimizzare VAE con variabili latenti discrete."]} +{"source": "Distributed computing can significantly reduce the training time of neural networks. Despite its potential, however, distributed training has not been widely adopted: scaling the training process is difficult, and existing SGD methods require substantial tuning of hyperparameters and learning schedules to achieve sufficient accuracy when increasing the number of workers. In practice, such tuning can be prohibitively expensive given the huge number of potential hyperparameter configurations and the effort required to test each one. We propose DANA, a novel approach that scales out-of-the-box to large clusters using the same hyperparameters and learning schedule optimized for training on a single worker, while maintaining similar final accuracy without additional overhead. DANA estimates the future value of model parameters by adapting Nesterov Accelerated Gradient to a distributed setting, and so mitigates the effect of gradient staleness, one of the main difficulties in scaling SGD to more workers. Evaluation on three state-of-the-art network architectures and three datasets shows that DANA scales as well as or better than existing work without having to tune any hyperparameters or tweak the learning schedule. For example, DANA achieves 75.73% accuracy on ImageNet when training ResNet-50 with 16 workers, similar to the non-distributed baseline.", "target": ["Un nuovo algoritmo distribuito asincrono SGD che raggiunge un'accuratezza allo stato dell'arte sulle architetture esistenti senza alcun tuning aggiuntivo o overhead.", "Propone un miglioramento degli approcci ASGD esistenti a scaling di medie dimensioni utilizzando il momentum con SGD per il training asincrono attraverso un pool di lavoratori distribuiti.", "Questo articolo affronta il problema della staleness del gradiente contro le prestazioni parallele nel training distribuito del deep learning, e propone un approccio per stimare i futuri parametri del modello negli slave per ridurre gli effetti della latenza di comunicazione."]} +{"source": "This paper proposes a novel approach to train deep neural networks by unlocking the layer-wise dependency of backpropagation training. The approach employs additional modules called local critic networks besides the main network model to be trained, which are used to obtain error gradients without complete feedforward and backward propagation processes. We propose a cascaded learning strategy for these local networks. In addition, the approach is also useful from multi-model perspectives, including structural optimization of neural networks, computationally efficient progressive inference, and ensemble classification for performance improvement. Experimental results show the effectiveness of the proposed approach and suggest guidelines for determining appropriate algorithm parameters.", "target": ["Proponiamo un nuovo algoritmo di apprendimento delle deep neural network, che sblocca la dipendenza layer-wise della backpropagation.", "Un paradigma di training alternativo per le DNI in cui il modulo ausiliario è addestrato per approssimare direttamente l'output finale del modello originale, offrendo benefici collaterali.", "Descrive un metodo di training delle reti neurali senza blocco degli aggiornamenti."]} +{"source": "\\emph{Truncated Backpropagation Through Time} (truncated BPTT, \\cite{jaeger2002tutorial}) is a widespread method for learning recurrent computational graphs. Truncated BPTT keeps the computational benefits of \\emph{Backpropagation Through Time} (BPTT \\cite{werbos:bptt}) while relieving the need for a complete backtrack through the whole data sequence at every step. However, truncation favors short-term dependencies: the gradient estimate of truncated BPTT is biased, so that it does not benefit from the convergence guarantees from stochastic gradient theory. We introduce \\emph{Anticipated Reweighted Truncated Backpropagation} (ARTBP), an algorithm that keeps the computational benefits of truncated BPTT, while providing unbiasedness. ARTBP works by using variable truncation lengths together with carefully chosen compensation factors in the backpropagation equation. We check the viability of ARTBP on two tasks. First, a simple synthetic task where careful balancing of temporal dependencies at different scales is needed: truncated BPTT displays unreliable performance, and in worst case scenarios, divergence, while ARTBP converges reliably. Second, on Penn Treebank character-level language modelling \\cite{ptb_proc}, ARTBP slightly outperforms truncated BPTT.", "target": ["Fornisce una versione unbiased della backpropagation troncata campionando le lunghezze di troncamento e ripesando di conseguenza.", "Propone metodi di determinazione stocastica per i punti di troncamento nella backpropagation attraverso il tempo.", "Una nuova approssimazione alla backpropagation attraverso il tempo per superare i carichi computazionali e di memoria che sorgono quando si deve imparare da lunghe sequenze."]} +{"source": "Graph convolutional networks (GCNs) have been widely used for classifying graph nodes in the semi-supervised setting. Previous works have shown that GCNs are vulnerable to the perturbation on adjacency and feature matrices of existing nodes. However, it is unrealistic to change the connections of existing nodes in many applications, such as existing users in social networks. In this paper, we investigate methods attacking GCNs by adding fake nodes. A greedy algorithm is proposed to generate adjacency and feature matrices of fake nodes, aiming to minimize the classification accuracy on the existing ones. In additional, we introduce a discriminator to classify fake nodes from real nodes, and propose a Greedy-GAN algorithm to simultaneously update the discriminator and the attacker, to make fake nodes indistinguishable to the real ones. Our non-targeted attack decreases the accuracy of GCN down to 0.10, and our targeted attack reaches a success rate of 0.99 for attacking the whole datasets, and 0.94 on average for attacking a single node.", "target": ["Attacco non mirato e mirato su GCN aggiungendo nodi falsi", "Gli autori propongono una nuova tecnica adversarial per aggiungere nodi \"falsi\" per ingannare un classificatore basato su GCN"]} +{"source": "Transfer learning aims to solve the data sparsity for a specific domain by applying information of another domain. Given a sequence (e.g. a natural language sentence), the transfer learning, usually enabled by recurrent neural network (RNN), represent the sequential information transfer. RNN uses a chain of repeating cells to model the sequence data. However, previous studies of neural network based transfer learning simply transfer the information across the whole layers, which are unfeasible for seq2seq and sequence labeling. Meanwhile, such layer-wise transfer learning mechanisms also lose the fine-grained cell-level information from the source domain. In this paper, we proposed the aligned recurrent transfer, ART, to achieve cell-level information transfer. ART is in a recurrent manner that different cells share the same parameters. Besides transferring the corresponding information at the same position, ART transfers information from all collocated words in the source domain. This strategy enables ART to capture the word collocation across domains in a more flexible way. We conducted extensive experiments on both sequence labeling tasks (POS tagging, NER) and sentence classification (sentiment analysis). ART outperforms the state-of-the-arts over all experiments.", "target": ["Transfer learning per sequenze tramite l'apprendimento per allineare le informazioni a livello di cell attraverso i domini.", "L'articolo ha proposto di usare RNN/LSTM con allineamento di collocazione come un metodo di representation learning per transfer learning/domain adaptation in NLP."]} +{"source": "Addressing uncertainty is critical for autonomous systems to robustly adapt to the real world. We formulate the problem of model uncertainty as a continuous Bayes-Adaptive Markov Decision Process (BAMDP), where an agent maintains a posterior distribution over latent model parameters given a history of observations and maximizes its expected long-term reward with respect to this belief distribution. Our algorithm, Bayesian Policy Optimization, builds on recent policy optimization algorithms to learn a universal policy that navigates the exploration-exploitation trade-off to maximize the Bayesian value function. To address challenges from discretizing the continuous latent parameter space, we propose a new policy network architecture that encodes the belief distribution independently from the observable state. Our method significantly outperforms algorithms that address model uncertainty without explicitly reasoning about belief distributions and is competitive with state-of-the-art Partially Observable Markov Decision Process solvers.", "target": ["Formuliamo l'incertezza del modello nel Reinforcement Learning come un processo decisionale continuo di Bayes-Adaptive Markov e presentiamo un metodo per l'ottimizzazione pratica e scalabile della policy bayesiana.", "Usando un approccio bayesiano, c'è un migliore compromesso tra exploration e exploitation nel RL"]} +{"source": "For many evaluation metrics commonly used as benchmarks for unconditional image generation, trivially memorizing the training set attains a better score than models which are considered state-of-the-art; we consider this problematic. We clarify a necessary condition for an evaluation metric not to behave this way: estimating the function must require a large sample from the model. In search of such a metric, we turn to neural network divergences (NNDs), which are defined in terms of a neural network trained to distinguish between distributions. The resulting benchmarks cannot be ``won'' by training set memorization, while still being perceptually correlated and computable only from samples. We survey past work on using NNDs for evaluation, implement an example black-box metric based on these ideas, and validate experimentally that it can measure a notion of generalization.", "target": ["Noi sosteniamo che i benchmark GAN devono richiedere un grande sample dal modello per penalizzare la memorizzazione e indagare se le divergenze delle reti neurali hanno questa proprietà.", "Gli autori propongono un criterio per valutare la qualità dei sample prodotti da una Generative Adversarial Network."]} +{"source": "Conventional methods model open domain dialogue generation as a black box through end-to-end learning from large scale conversation data. In this work, we make the first step to open the black box by introducing dialogue acts into open domain dialogue generation. The dialogue acts are generally designed and reveal how people engage in social chat. Inspired by analysis on real data, we propose jointly modeling dialogue act selection and response generation, and perform learning with human-human conversations tagged with a dialogue act classifier and a reinforcement approach to further optimizing the model for long-term conversation. With the dialogue acts, we not only achieve significant improvement over state-of-the-art methods on response quality for given contexts and long-term conversation in both machine-machine simulation and human-machine conversation, but also are capable of explaining why such achievements can be made.", "target": ["Generazione di dialogo open-domain con atti di dialogo", "Gli autori usano una tecnica di distant supervision per aggiungere tag di atti di dialogo come fattore condizionante per generare risposte nei dialoghi open-domain", "L'articolo descrive una tecnica per incorporare atti di dialogo in agenti di conversazione neurali"]} +{"source": "We discuss the feasibility of the following learning problem: given unmatched samples from two domains and nothing else, learn a mapping between the two, which preserves semantics. Due to the lack of paired samples and without any definition of the semantic information, the problem might seem ill-posed. Specifically, in typical cases, it seems possible to build infinitely many alternative mappings from every target mapping. This apparent ambiguity stands in sharp contrast to the recent empirical success in solving this problem. We identify the abstract notion of aligning two domains in a semantic way with concrete terms of minimal relative complexity. A theoretical framework for measuring the complexity of compositions of functions is developed in order to show that it is reasonable to expect the minimal complexity mapping to be unique. The measured complexity used is directly related to the depth of the neural networks being learned and a semantically aligned mapping could then be captured simply by learning using architectures that are not much bigger than the minimal architecture. Various predictions are made based on the hypothesis that semantic alignment can be captured by the minimal mapping. These are verified extensively. In addition, a new mapping algorithm is proposed and shown to lead to better mapping results.", "target": ["La nostra ipotesi è che, dati due domini, la mappatura di minor complessità che ha una bassa discrepanza approssima la mappatura di destinazione.", "L'articolo affronta il problema dell'apprendimento di mappature tra diversi domini senza alcuna supervisione, affermando tre congetture.", "Dimostra che nell'apprendimento unsupervised su dati non allineati è possibile imparare la mappatura tra domini usando solo GAN senza una loss di ricostruzione."]} +{"source": "We present a novel approach for the certification of neural networks against adversarial perturbations which combines scalable overapproximation methods with precise (mixed integer) linear programming. This results in significantly better precision than state-of-the-art verifiers on challenging feedforward and convolutional neural networks with piecewise linear activation functions.", "target": ["Raffiniamo i risultati di sovra-approssimazione dei verificatori incompleti usando solutori MILP per dimostrare proprietà di robustezza maggiori rispetto allo stato dell'arte.", "Introduce un verificatore che ottiene un miglioramento della precisione dei verificatori incompleti e della scalabilità dei verificatori completi usando la sovra-parametrizzazione, la programmazione lineare intera mista e il rilassamento della programmazione lineare.", "Una strategia mista per ottenere una migliore precisione nelle verifiche di robustezza delle reti neurali feed-forward con funzioni di attivazione lineari piecewise, ottenendo una migliore precisione rispetto ai verificatori incompleti e una maggiore scalabilità rispetto ai verificatori completi."]} +{"source": "A distinct commonality between HMMs and RNNs is that they both learn hidden representations for sequential data. In addition, it has been noted that the backward computation of the Baum-Welch algorithm for HMMs is a special case of the back-propagation algorithm used for neural networks (Eisner (2016)). Do these observations suggest that, despite their many apparent differences, HMMs are a special case of RNNs? In this paper, we show that that is indeed the case, and investigate a series of architectural transformations between HMMs and RNNs, both through theoretical derivations and empirical hybridization. In particular, we investigate three key design factors—independence assumptions between the hidden states and the observation, the placement of softmaxes, and the use of non-linearities—in order to pin down their empirical effects. We present a comprehensive empirical study to provide insights into the interplay between expressivity and interpretability in this model family with respect to language modeling and parts-of-speech induction.", "target": ["Gli HMM sono un caso speciale di RNN? Indaghiamo una serie di trasformazioni dell'architettura tra HMM e RNN, sia attraverso derivazioni teoriche che ibridazioni empiriche e forniamo nuove intuizioni.", "Questo articolo esplora se gli HMM sono un caso speciale di RNN utilizzando il language modeling e il POS tagging"]} +{"source": "Deep neural networks have been tremendously successful in a number of tasks. One of the main reasons for this is their capability to automatically learn representations of data in levels of abstraction, increasingly disentangling the data as the internal transformations are applied. In this paper we propose a novel regularization method that penalize covariance between dimensions of the hidden layers in a network, something that benefits the disentanglement. This makes the network learn nonlinear representations that are linearly uncorrelated, yet allows the model to obtain good results on a number of tasks, as demonstrated by our experimental evaluation. The proposed technique can be used to find the dimensionality of the underlying data, because it effectively disables dimensions that aren't needed. Our approach is simple and computationally cheap, as it can be applied as a regularizer to any gradient-based learning model.", "target": ["Proponiamo un nuovo metodo di regolarizzazione che penalizza la covarianza tra le dimensioni degli hidden layer in una rete.", "Questo articolo presenta un meccanismo di regolarizzazione che penalizza la covarianza tra tutte le dimensioni nella rappresentazione latente di una rete neurale al fine di dissociare la rappresentazione latente"]} +{"source": "This report introduces a training and recognition scheme, in which classification is realized via class-wise discerning. Trained with datasets whose labels are randomly shuffled except for one class of interest, a neural network learns class-wise parameter values, and remolds itself from a feature sorter into feature filters, each of which discerns objects belonging to one of the classes only. Classification of an input can be inferred from the maximum response of the filters. A multiple check with multiple versions of filters can diminish fluctuation and yields better performance. This scheme of discerning, maximum response and multiple check is a method of general viability to improve performance of feedforward networks, and the filter training itself is a promising feature abstraction procedure. In contrast to the direct sorting, the scheme mimics the classification process mediated by a series of one component picking.", "target": ["Lo schema proposto imita il processo di classificazione mediato da una serie di prelievi di un componente.", "Un metodo per aumentare l'accuratezza delle deep network su task di classificazione multiclasse tramite una riduzione della classificazione multiclasse a quella binaria.", "Una nuova procedura di classificazione per il discernimento, la risposta massima e il controllo multiplo per migliorare la precisione delle reti mediocri e migliorare le reti feedforward."]} +{"source": "A long-held conventional wisdom states that larger models train more slowly when using gradient descent. This work challenges this widely-held belief, showing that larger models can potentially train faster despite the increasing computational requirements of each training step. In particular, we study the effect of network structure (depth and width) on halting time and show that larger models---wider models in particular---take fewer training steps to converge. We design simple experiments to quantitatively characterize the effect of overparametrization on weight space traversal. Results show that halting time improves when growing model's width for three different applications, and the improvement comes from each factor: The distance from initialized weights to converged weights shrinks with a power-law-like relationship, the average step size grows with a power-law-like relationship, and gradient vectors become more aligned with each other during traversal.", "target": ["Empiricamente dimostra che i modelli più grandi si addestrano in meno step di allenamento, perché tutti i fattori di attraversamento dello spazio dei pesi migliorano.", "Questo articolo mostra che RNN più larghe migliorano la velocità di convergenza quando sono applicate a problemi di NLP, e per estensione mostra anche l'effetto di aumentare le larghezze nelle deep neural network sulla convergenza dell'ottimizzazione", "Questo articolo caratterizza l'impatto della sovra-parametrizzazione nel numero di iterazioni necessarie a un algoritmo per convergere, e presenta ulteriori osservazioni empiriche sugli effetti della sovra-parametrizzazione nel training delle reti neurali."]} +{"source": "Due to its potential to improve programmer productivity and software quality, automated program repair has been an active topic of research. Newer techniques harness neural networks to learn directly from examples of buggy programs and their fixes. In this work, we consider a recently identified class of bugs called variable-misuse bugs. The state-of-the-art solution for variable misuse enumerates potential fixes for all possible bug locations in a program, before selecting the best prediction. We show that it is beneficial to train a model that jointly and directly localizes and repairs variable-misuse bugs. We present multi-headed pointer networks for this purpose, with one head each for localization and repair. The experimental results show that the joint model significantly outperforms an enumerative solution that uses a pointer based model for repair alone.", "target": ["Multi-headed Pointer Network per apprendere congiuntamente a localizzare e a riparare i bug di abuso di variabili", "Propone un modello basato su LSTM con puntatori per suddividere il problema del VarMisuse in più step.", "Questo articolo presenta un modello basato su LSTM per il rilevamento e la riparazione dei bug di VarMisuse, e dimostra miglioramenti significativi rispetto agli approcci precedenti su diversi dataset."]} +{"source": "Classification and clustering have been studied separately in machine learning and computer vision. Inspired by the recent success of deep learning models in solving various vision problems (e.g., object recognition, semantic segmentation) and the fact that humans serve as the gold standard in assessing clustering algorithms, here, we advocate for a unified treatment of the two problems and suggest that hierarchical frameworks that progressively build complex patterns on top of the simpler ones (e.g., convolutional neural networks) offer a promising solution. We do not dwell much on the learning mechanisms in these frameworks as they are still a matter of debate, with respect to biological constraints. Instead, we emphasize on the compositionality of the real world structures and objects. In particular, we show that CNNs, trained end to end using back propagation with noisy labels, are able to cluster data points belonging to several overlapping shapes, and do so much better than the state of the art algorithms. The main takeaway lesson from our study is that mechanisms of human vision, particularly the hierarchal organization of the visual ventral stream should be taken into account in clustering algorithms (e.g., for learning representations in an unsupervised manner or with minimum supervision) to reach human level clustering performance. This, by no means, suggests that other methods do not hold merits. For example, methods relying on pairwise affinities (e.g., spectral clustering) have been very successful in many cases but still fail in some cases (e.g., overlapping clusters).", "target": ["Clustering di tipo umano con CNN", "L'articolo convalida l'idea che le reti neurali convoluzionali profonde potrebbero imparare a raggruppare i dati di input meglio di altri metodi di clustering notando la loro capacità di interpretare il contesto di ogni punto di input grazie a un ampio campo visivo.", "Questo lavoro combina il deep learning per la rappresentazione delle feature con il task di clustering unsupervised simile a quello umano."]} +{"source": "Instancewise feature scoring is a method for model interpretation, which yields, for each test instance, a vector of importance scores associated with features. Methods based on the Shapley score have been proposed as a fair way of computing feature attributions, but incur an exponential complexity in the number of features. This combinatorial explosion arises from the definition of Shapley value and prevents these methods from being scalable to large data sets and complex models. We focus on settings in which the data have a graph structure, and the contribution of features to the target variable is well-approximated by a graph-structured factorization. In such settings, we develop two algorithms with linear complexity for instancewise feature importance scoring on black-box models. We establish the relationship of our methods to the Shapley value and a closely related concept known as the Myerson value from cooperative game theory. We demonstrate on both language and image data that our algorithms compare favorably with other methods using both quantitative metrics and human evaluation.", "target": ["Sviluppiamo due algoritmi di complessità lineare per l'interpretazione di modelli model-agnostic basata sul valore di Shapley, nel setting in cui il contributo delle feature all'obiettivo è ben approssimato da una fattorizzazione strutturata a grafo.", "L'articolo propone due approssimazioni al valore di Shapley usato per generare i punteggi delle feature per l'interpretabilità.", "Questo articolo propone due metodi per la valutazione dell'importanza delle feature in base all'istanza utilizzando i valori di Shapely, e fornisce due metodi efficienti per calcolare i valori di Shapely approssimati quando c'è una struttura nota che mette in relazione le feature."]} +{"source": "According to parallel distributed processing (PDP) theory in psychology, neural networks (NN) learn distributed rather than interpretable localist representations. This view has been held so strongly that few researchers have analysed single units to determine if this assumption is correct. However, recent results from psychology, neuroscience and computer science have shown the occasional existence of local codes emerging in artificial and biological neural networks. In this paper, we undertake the first systematic survey of when local codes emerge in a feed-forward neural network, using generated input and output data with known qualities. We find that the number of local codes that emerge from a NN follows a well-defined distribution across the number of hidden layer neurons, with a peak determined by the size of input data, number of examples presented and the sparsity of input data. Using a 1-hot output code drastically decreases the number of local codes on the hidden layer. The number of emergent local codes increases with the percentage of dropout applied to the hidden layer, suggesting that the localist encoding may offer a resilience to noisy networks. This data suggests that localist coding can emerge from feed-forward PDP networks and suggests some of the conditions that may lead to interpretable localist representations in the cortex. The findings highlight how local codes should not be dismissed out of hand.", "target": ["Sono stati trovati codici locali nelle reti neurali feed-forward", "Un metodo per determinare fino a che punto i singoli neuroni in un hidden layer di una MLP codifichino un codice localista, che viene studiato per diverse rappresentazioni di input.", "Studia lo sviluppo di rappresentazioni localistiche negli hidden layer delle reti neurali feed-forward."]} +{"source": "Representing entities and relations in an embedding space is a well-studied approach for machine learning on relational data. Existing approaches however primarily focus on simple link structure between a finite set of entities, ignoring the variety of data types that are often used in relational databases, such as text, images, and numerical values. In our approach, we propose a multimodal embedding using different neural encoders for this variety of data, and combine with existing models to learn embeddings of the entities. We extend existing datasets to create two novel benchmarks, YAGO-10-plus and MovieLens-100k-plus, that contain additional relations such as textual descriptions and images of the original entities. We demonstrate that our model utilizes the additional information effectively to provide further gains in accuracy. Moreover, we test our learned multimodal embeddings by using them to predict missing multimodal attributes.", "target": ["Estensione della modellazione relazionale per supportare i dati multimodali usando codificatori neurali.", "Questo articolo propone di eseguire la predizione dei link nelle knowledge base integrando le entità originali con informazioni multimodali, e presenta un modello in grado di codificare tutti i tipi di informazioni quando assegna un punteggio alle triple.", "L'articolo riguarda l'embedding di informazioni di diversa natura negli approcci di link prediction"]} +{"source": "An ensemble of neural networks is known to be more robust and accurate than an individual network, however usually with linearly-increased cost in both training and testing. In this work, we propose a two-stage method to learn Sparse Structured Ensembles (SSEs) for neural networks. In the first stage, we run SG-MCMC with group sparse priors to draw an ensemble of samples from the posterior distribution of network parameters. In the second stage, we apply weight-pruning to each sampled network and then perform retraining over the remained connections. In this way of learning SSEs with SG-MCMC and pruning, we not only achieve high prediction accuracy since SG-MCMC enhances exploration of the model-parameter space, but also reduce memory and computation cost significantly in both training and testing of NN ensembles. This is thoroughly evaluated in the experiments of learning SSE ensembles of both FNNs and LSTMs. For example, in LSTM based language modeling (LM), we obtain 21\\% relative reduction in LM perplexity by learning a SSE of 4 large LSTM models, which has only 30\\% of model parameters and 70\\% of computations in total, as compared to the baseline large LSTM LM. To the best of our knowledge, this work represents the first methodology and empirical study of integrating SG-MCMC, group sparse prior and network pruning together for learning NN ensembles.", "target": ["Si propone un nuovo metodo integrando il sampling SG-MCMC, il prior sparso di gruppo e il pruning della rete per imparare Sparse Structured Ensemble (SSE) con prestazioni migliori e costi significativamente ridotti rispetto ai metodi tradizionali.", "Gli autori propongono una procedura per generare un'ensemble di modelli strutturati sparsi", "Un nuovo framework per il training di ensemble di reti neurali che utilizza metodi SG-MCMC all'interno del deep learning, e poi aumenta l'efficienza computazionale tramite sparsità+pruning di gruppo.", "Questo articolo esplora l'uso di FNN e LSTM per rendere la media dei modelli bayesiani più fattibile da computare e per migliorare le prestazioni medie dei modelli."]} +{"source": "This paper introduces a new framework for data efficient and versatile learning. Specifically: 1) We develop ML-PIP, a general framework for Meta-Learning approximate Probabilistic Inference for Prediction. ML-PIP extends existing probabilistic interpretations of meta-learning to cover a broad class of methods. 2) We introduce \\Versa{}, an instance of the framework employing a flexible and versatile amortization network that takes few-shot learning datasets as inputs, with arbitrary numbers of shots, and outputs a distribution over task-specific parameters in a single forward pass. \\Versa{} substitutes optimization at test time with forward passes through inference networks, amortizing the cost of inference and relieving the need for second derivatives during training. 3) We evaluate \\Versa{} on benchmark datasets where the method sets new state-of-the-art results, and can handle arbitrary number of shots, and for classification, arbitrary numbers of classes at train and test time. The power of the approach is then demonstrated through a challenging few-shot ShapeNet view reconstruction task.", "target": ["Nuovo framework per il meta-learning che unifica ed estende un'ampia classe di metodi esistenti per il few-shot learning. Raggiunge una forte performance sui benchmark di few-shot learning senza richiedere un'inferenza iterativa al test-time.", "Questo lavoro affronta il few-shot learning da un punto di vista dell'inferenza probabilistica, raggiungendo lo stato dell'arte nonostante una configurazione più semplice di molti altri concorrenti"]} +{"source": "In recent years, softmax together with its fast approximations has become the de-facto loss function for deep neural networks with multiclass predictions. However, softmax is used in many problems that do not fully fit the multiclass framework and where the softmax assumption of mutually exclusive outcomes can lead to biased results. This is often the case for applications such as language modeling, next event prediction and matrix factorization, where many of the potential outcomes are not mutually exclusive, but are more likely to be independent conditionally on the state. To this end, for the set of problems with positive and unlabeled data, we propose a relaxation of the original softmax formulation, where, given the observed state, each of the outcomes are conditionally independent but share a common set of negatives. Since we operate in a regime where explicit negatives are missing, we create an adversarially-trained model of negatives and derive a new negative sampling and weighting scheme which we denote as Cooperative Importance Sampling (CIS). We show empirically the advantages of our newly introduced negative sampling scheme by pluging it in the Word2Vec algorithm and benching it extensively against other negative sampling schemes on both language modeling and matrix factorization tasks and show large lifts in performance.", "target": ["Definizione di una loss softmax parzialmente mutualmente esclusiva per dati positivi e implementazione di uno schema di sampling basato sulla cooperazione", "Questo articolo presenta il Cooperative Importance Sampling per risolvere il problema dell'assunzione reciprocamente esclusiva del softmax tradizionale che è biased quando i sample negativi non sono esplicitamente definiti", "Questo articolo propone metodi PMES per rilassare l'ipotesi di risultato esclusivo nella loss softmax, dimostrandone il merito empirico nel migliorare i modelli di embedding di tipo word2vec."]} +{"source": "Over the past few years, various tasks involving videos such as classification, description, summarization and question answering have received a lot of attention. Current models for these tasks compute an encoding of the video by treating it as a sequence of images and going over every image in the sequence, which becomes computationally expensive for longer videos. In this paper, we focus on the task of video classification and aim to reduce the computational cost by using the idea of distillation. Specifically, we propose a Teacher-Student network wherein the teacher looks at all the frames in the video but the student looks at only a small fraction of the frames in the video. The idea is to then train the student to minimize (i) the difference between the final representation computed by the student and the teacher and/or (ii) the difference between the distributions predicted by the teacher and the student. This smaller student network which involves fewer computations but still learns to mimic the teacher can then be employed at inference time for video classification. We experiment with the YouTube-8M dataset and show that the proposed student network can reduce the inference time by upto 30% with a negligent drop in the performance.", "target": ["Struttura teacher-student per una classificazione video efficiente utilizzando un minor numero di frame", "L'articolo propone un'idea per distillare da un modello di classificazione video completo un piccolo modello che riceve solo un numero minore di frame.", "Gli autori presentano una rete teacher-student per risolvere il problema della classificazione dei video, proponendo algoritmi di training seriali e paralleli volti a ridurre i costi computazionali."]} +{"source": "Deep generative models have achieved impressive success in recent years. Generative Adversarial Networks (GANs) and Variational Autoencoders (VAEs), as powerful frameworks for deep generative model learning, have largely been considered as two distinct paradigms and received extensive independent studies respectively. This paper aims to establish formal connections between GANs and VAEs through a new formulation of them. We interpret sample generation in GANs as performing posterior inference, and show that GANs and VAEs involve minimizing KL divergences of respective posterior and inference distributions with opposite directions, extending the two learning phases of classic wake-sleep algorithm, respectively. The unified view provides a powerful tool to analyze a diverse set of existing model variants, and enables to transfer techniques across research lines in a principled way. For example, we apply the importance weighting method in VAE literatures for improved GAN learning, and enhance VAEs with an adversarial mechanism that leverages generated samples. Experiments show generality and effectiveness of the transfered techniques.", "target": ["Una visione statistica unificata dell'ampia classe di deep generative model", "L'articolo sviluppa un framework che interpreta gli algoritmi GAN come l'esecuzione di una forma di inferenza variazionale su un modello generativo che ricostruisce una variabile indicatrice del fatto che un sample appartenga alla vera distribuzione di dati generativi."]} +{"source": "Deep neural networks have demonstrated promising prediction and classification performance on many healthcare applications. However, the interpretability of those models are often lacking. On the other hand, classical interpretable models such as rule lists or decision trees do not lead to the same level of accuracy as deep neural networks and can often be too complex to interpret (due to the potentially large depth of rule lists). In this work, we present PEARL, Prototype lEArning via Rule Lists, which iteratively uses rule lists to guide a neural network to learn representative data prototypes. The resulting prototype neural network provides accurate prediction, and the prediction can be easily explained by prototype and its guiding rule lists. Thanks to the prediction power of neural networks, the rule lists from\t\t\t\t prototypes are more concise and hence provide better interpretability. On two real-world electronic healthcare records (EHR) datasets, PEARL consistently outperforms all baselines across both datasets, especially achieving performance improvement over conventional rule learning by up to 28% and over prototype learning by up to 3%. Experimental results also show the resulting interpretation of PEARL is simpler than the standard rule learning.", "target": ["Un metodo che combina l'apprendimento di liste di regole e il prototype learning", "Presenta un nuovo framework di predizione interpretabile, che combina l'apprendimento basato su regole, il prototype learning e le NN ed è particolarmente applicabile ai dati longitudinali.", "Questo articolo mira ad affrontare la mancanza di interpretabilità dei modelli di deep learning, e propone Prototype Learning via Rule Lists (PEARL), che combina l'apprendimento delle regole e prototype learning per ottenere una classificazione più accurata e rendere il task di interpretabilità più semplice."]} +{"source": "Generative Adversarial Networks (GANs) are powerful tools for realistic image generation. However, a major drawback of GANs is that they are especially hard to train, often requiring large amounts of data and long training time. In this paper we propose the Deli-Fisher GAN, a GAN that generates photo-realistic images by enforcing structure on the latent generative space using similar approaches in \\cite{deligan}. The structure of the latent space we consider in this paper is modeled as a mixture of Gaussians, whose parameters are learned in the training process. Furthermore, to improve stability and efficiency, we use the Fisher Integral Probability Metric as the divergence measure in our GAN model, instead of the Jensen-Shannon divergence. We show by experiments that the Deli-Fisher GAN performs better than DCGAN, WGAN, and the Fisher GAN as measured by inception score.", "target": ["Questo articolo propone una nuova Generative Adversarial Network che è più stabile, più efficiente e produce immagini migliori di quelle dello status-quo", "Questo paper combina Fisher-GAN e Deli-GAN", "Questo paper combina Deli-GAN, che ha una distribuzione a priori mista nello spazio latente, e Fisher GAN, che usa Fisher IPM invece di JSD come obiettivo."]} +{"source": "Recent work on encoder-decoder models for sequence-to-sequence mapping has shown that integrating both temporal and spatial attentional mechanisms into neural networks increases the performance of the system substantially. We report on a new modular network architecture that applies an attentional mechanism not on temporal and spatial regions of the input, but on sensor selection for multi-sensor setups. This network called the sensor transformation attention network (STAN) is evaluated in scenarios which include the presence of natural noise or synthetic dynamic noise. We demonstrate how the attentional signal responds dynamically to changing noise levels and sensor-specific noise, leading to reduced word error rates (WERs) on both audio and visual tasks using TIDIGITS and GRID; and also on CHiME-3, a multi-microphone real-world noisy dataset. The improvement grows as more channels are corrupted as demonstrated on the CHiME-3 dataset. Moreover, the proposed STAN architecture naturally introduces a number of advantages including ease of removing sensors from existing architectures, attentional interpretability, and increased robustness to a variety of noise environments.", "target": ["Introduciamo un'architettura di rete modulare multisensore con un meccanismo di attention che permette la selezione dinamica del sensore su dati rumorosi del mondo reale da CHiME-3.", "Un'architettura neurale generica in grado di apprendere l'attention che deve essere prestata ai diversi canali di input a seconda della qualità relativa di ogni sensore rispetto agli altri.", "Considera l'uso dell'attention per la selezione del sensore o del canale con risultati su TIDIGITS e GRID che mostrano un beneficio dell'attention rispetto alla concatenazione delle feature."]} +{"source": "Massive data exist among user local platforms that usually cannot support deep neural network (DNN) training due to computation and storage resource constraints. Cloud-based training schemes provide beneficial services but suffer from potential privacy risks due to excessive user data collection. To enable cloud-based DNN training while protecting the data privacy simultaneously, we propose to leverage the intermediate representations of the data, which is achieved by splitting the DNNs and deploying them separately onto local platforms and the cloud. The local neural network (NN) is used to generate the feature representations. To avoid local training and protect data privacy, the local NN is derived from pre-trained NNs. The cloud NN is then trained based on the extracted intermediate representations for the target learning task. We validate the idea of DNN splitting by characterizing the dependency of privacy loss and classification accuracy on the local NN topology for a convolutional NN (CNN) based image classification task. Based on the characterization, we further propose PrivyNet to determine the local NN topology, which optimizes the accuracy of the target learning task under the constraints on privacy loss, local computation, and storage. The efficiency and effectiveness of PrivyNet are demonstrated with CIFAR-10 dataset.", "target": ["Per consentire il training DNN basato sul cloud e proteggere contemporaneamente la privacy dei dati, proponiamo di sfruttare le rappresentazioni intermedie dei dati, il che si ottiene dividendo le DNN e distribuendole separatamente su piattaforme locali e sul cloud.", "Questo articolo propone una tecnica per privatizzare i dati imparando una rappresentazione delle feature che è difficile da usare per la ricostruzione delle immagini, ma utile per la classificazione delle stesse."]} +{"source": "Generative Adversarial Networks (GANs) have shown remarkable success as a framework for training models to produce realistic-looking data. In this work, we propose a Recurrent GAN (RGAN) and Recurrent Conditional GAN (RCGAN) to produce realistic real-valued multi-dimensional time series, with an emphasis on their application to medical data. RGANs make use of recurrent neural networks (RNNs) in the generator and the discriminator. In the case of RCGANs, both of these RNNs are conditioned on auxiliary information. We demonstrate our models in a set of toy datasets, where we show visually and quantitatively (using sample likelihood and maximum mean discrepancy) that they can successfully generate realistic time-series. We also describe novel evaluation methods for GANs, where we generate a synthetic labelled training dataset, and evaluate on a real test set the performance of a model trained on the synthetic data, and vice-versa. We illustrate with these metrics that RCGANs can generate time-series data useful for supervised training, with only minor degradation in performance on real test data. This is demonstrated on digit classification from ‘serialised’ MNIST and by training an early warning system on a medical dataset of 17,000 patients from an intensive care unit. We further discuss and analyse the privacy concerns that may arise when using RCGANs to generate realistic synthetic medical time series data, and demonstrate results from differentially private training of the RCGAN.", "target": ["GAN ricorrenti condizionali per la generazione di sequenze mediche a valori reali, mostrando nuovi approcci di valutazione e un'analisi empirica della privacy.", "Propone di utilizzare i dati sintetici generati da GAN come sostituzione dei dati personali identificabili nel training di modelli ML per applicazioni sensibili alla privacy", "Gli autori propongono una nuova architettura GAN ricorrente che genera sequenze di dominio continue, e la valutano su diversi task sintetici e su un task con dati costituiti da serie temporali ICU.", "Propone di utilizzare RGAN e RCGAN per generare sequenze sintetiche di dati reali."]} +{"source": "Emphasis effects – visual changes that make certain elements more prominent – are commonly used in information visualization to draw the user’s attention or to indicate importance. Although theoretical frameworks of emphasis exist (that link visually diverse emphasis effects through the idea of visual prominence compared to background elements), most metrics for predicting how emphasis effects will be perceived by users come from abstract models of human vision which may not apply to visualization design. In particular, it is difficult for designers to know, when designing a visualization, how different emphasis effects will compare and what level of one effect is equivalent to what level of another. To address this gap, we carried out two studies that provide empirical evidence about how users perceive different emphasis effects, using three visual variables (colour, size, and blur/focus) and eight strength levels. Results from gaze tracking, mouse clicks, and subjective responses show that there are significant differences between visual variables and between levels, and allow us to develop an initial understanding of perceptual equivalence. We developed a model from the data in our first study, and used it to predict the results in the second; the model was accurate, with high correlations between predictions and real values. Our studies and empirical models provide valuable new information for designers who want to understand and control how emphasis effects will be perceived by users.", "target": ["I nostri studi e modelli empirici forniscono nuove informazioni preziose per i designer che vogliono capire e controllare come gli effetti dell'enfasi saranno percepiti dagli utenti", "Questo articolo considera quale evidenziazione visiva venga percepita più velocemente nella visualizzazione dei dati e come i diversi metodi di evidenziazione si confrontino tra loro", "Due studi sull'efficacia degli effetti di enfasi, uno che valuta i livelli di differenze utili, e uno più applicato utilizzando visualizzazioni effettive diverse per un'indagine più ecologicamente valida."]} +{"source": "Memory Network based models have shown a remarkable progress on the task of relational reasoning. Recently, a simpler yet powerful neural network module called Relation Network (RN) has been introduced. Despite its architectural simplicity, the time complexity of relation network grows quadratically with data, hence limiting its application to tasks with a large-scaled memory. We introduce Related Memory Network, an end-to-end neural network architecture exploiting both memory network and relation network structures. We follow memory network's four components while each component operates similar to the relation network without taking a pair of objects. As a result, our model is as simple as RN but the computational complexity is reduced to linear time. It achieves the state-of-the-art results in jointly trained bAbI-10k story-based question answering and bAbI dialog dataset.", "target": ["Una semplice architettura di ragionamento basata sulle memory network (MemNN) e la relation network (RN), riducendo la complessità temporale rispetto alla RN e raggiungendo un risultato allo stato dell'arte sulla storia bAbI basata su QA e sul dialogo bAbI.", "Introduce la Related memory network (RMN), un miglioramento rispetto alle relation network (RN)."]} +{"source": "We investigate in this paper the architecture of deep convolutional networks. Building on existing state of the art models, we propose a reconfiguration of the model parameters into several parallel branches at the global network level, with each branch being a standalone CNN. We show that this arrangement is an efficient way to significantly reduce the number of parameters while at the same time improving the performance. The use of branches brings an additional form of regularization. In addition to splitting the parameters into parallel branches, we propose a tighter coupling of these branches by averaging their log-probabilities. The tighter coupling favours the learning of better representations, even at the level of the individual branches, as compared to when each branch is trained independently. We refer to this branched architecture as \"coupled ensembles\". The approach is very generic and can be applied with almost any neural network architecture. With coupled ensembles of DenseNet-BC and parameter budget of 25M, we obtain error rates of 2.92%, 15.68% and 1.50% respectively on CIFAR-10, CIFAR-100 and SVHN tasks. For the same parameter budget, DenseNet-BC has an error rate of 3.46%, 17.18%, and 1.8% respectively. With ensembles of coupled ensembles, of DenseNet-BC networks, with 50M total parameters, we obtain error rates of 2.72%, 15.13% and 1.42% respectively on these tasks.", "target": ["Mostriamo che dividere una rete neurale in rami paralleli migliora le prestazioni e che un corretto accoppiamento dei rami migliora ulteriormente le prestazioni.", "Il lavoro propone una riconfigurazione del modello CNN esistente allo stato dell'arte utilizzando una nuova architettura di ramificazione, con prestazioni migliori.", "Questo articolo mostra i benefici di risparmio di parametri dell'ensembling accoppiato.", "Presenta un'architettura di deep network che elabora i dati utilizzando più rami paralleli e combina il posteriore da questi rami per calcolare i punteggi finali."]} +{"source": "Convolutional Neural Networks (CNN) are very popular in many fields including computer vision, speech recognition, natural language processing, to name a few. Though deep learning leads to groundbreaking performance in these domains, the networks used are very demanding computationally and are far from real-time even on a GPU, which is not power efficient and therefore does not suit low power systems such as mobile devices. To overcome this challenge, some solutions have been proposed for quantizing the weights and activations of these networks, which accelerate the runtime significantly. Yet, this acceleration comes at the cost of a larger error. The NICE method proposed in this work trains quantized neural networks by noise injection and a learned clamping, which improve the accuracy. This leads to state-of-the-art results on various regression and classification tasks, e.g., ImageNet classification with architectures such as ResNet-18/34/50 with low as 3-bit weights and 3 -bit activations. We implement the proposed solution on an FPGA to demonstrate its applicability for low power real-time applications.", "target": ["Combina l'iniezione di rumore, la quantizzazione graduale e l'apprendimento del clamping dell'attivazione per ottenere una quantizzazione a 3, 4 e 5 bit allo stato dell'arte", "Propone di iniettare rumore durante il training e di bloccare i valori dei parametri in un layer così come l'output dell'attivazione nella quantizzazione della rete neurale.", "Un metodo per la quantizzazione delle deep neural network per la classificazione e la regressione, utilizzando l'iniezione di rumore, il clamping con attivazioni massime apprese e la quantizzazione graduale a blocchi per performare alla pari o meglio dei metodi allo stato dell'arte."]} +{"source": "In complex transfer learning scenarios new tasks might not be tightly linked to previous tasks. Approaches that transfer information contained only in the final parameters of a source model will therefore struggle. Instead, transfer learning at at higher level of abstraction is needed. We propose Leap, a framework that achieves this by transferring knowledge across learning processes. We associate each task with a manifold on which the training process travels from initialization to final parameters and construct a meta-learning objective that minimizes the expected length of this path. Our framework leverages only information obtained during training and can be computed on the fly at negligible cost. We demonstrate that our framework outperforms competing methods, both in meta-learning and transfer learning, on a set of computer vision tasks. Finally, we demonstrate that Leap can transfer knowledge across learning processes in demanding reinforcement learning environments (Atari) that involve millions of gradient steps.", "target": ["Proponiamo Leap, un framework che trasferisce la conoscenza tra i processi di apprendimento minimizzando la distanza prevista che il processo di training percorre sulla superficie della loss di un task.", "L'articolo propone un nuovo obiettivo di meta-learning volto a superare gli approcci allo stato dell'arte quando si tratta di collezioni di task che mostrano una sostanziale diversità tra i task"]} +{"source": "Given an existing trained neural network, it is often desirable to learn new capabilities without hindering performance of those already learned. Existing approaches either learn sub-optimal solutions, require joint training, or incur a substantial increment in the number of parameters for each added task, typically as many as the original network. We propose a method called Deep Adaptation Networks (DAN) that constrains newly learned filters to be linear combinations of existing ones. DANs preserve performance on the original task, require a fraction (typically 13%) of the number of parameters compared to standard fine-tuning procedures and converge in less cycles of training to a comparable or better level of performance. When coupled with standard network quantization techniques, we further reduce the parameter cost to around 3% of the original with negligible or no loss in accuracy. The learned architecture can be controlled to switch between various learned representations, enabling a single network to solve a task from multiple different domains. We conduct extensive experiments showing the effectiveness of our method on a range of image classification tasks and explore different aspects of its behavior.", "target": ["Un'alternativa al transfer learning che impara più velocemente, richiede molti meno parametri (3-13%), di solito raggiunge risultati migliori e conserva precisamente le prestazioni sui vecchi task.", "Moduli di controllo per l'apprendimento incrementale su dataset di classificazione delle immagini"]} +{"source": "High throughput and low latency inference of deep neural networks are critical for the deployment of deep learning applications. This paper presents a general technique toward 8-bit low precision inference of convolutional neural networks, including 1) channel-wise scale factors of weights, especially for depthwise convolution, 2) Winograd convolution, and 3) topology-wise 8-bit support. We experiment the techniques on top of a widely-used deep learning framework. The 8-bit optimized model is automatically generated with a calibration process from FP32 model without the need of fine-tuning or retraining. We perform a systematical and comprehensive study on 18 widely-used convolutional neural networks and demonstrate the effectiveness of 8-bit low precision inference across a wide range of applications and use cases, including image classification, object detection, image segmentation, and super resolution. We show that the inference throughput and latency are improved by 1.6X and 1.5X respectively with minimal within 0.6%1to no loss in accuracy from FP32 baseline. We believe the methodology can provide the guidance and reference design of 8-bit low precision inference for other frameworks. All the code and models will be publicly available soon.", "target": ["Presentiamo una tecnica generale per l'inferenza a 8 bit a bassa precisione delle reti neurali convoluzionali.", "Questo articolo progetta un sistema per quantizzare automaticamente i modelli preaddestrati CNN"]} +{"source": "Recent approaches have successfully demonstrated the benefits of learning the parameters of shallow networks in hyperbolic space. We extend this line of work by imposing hyperbolic geometry on the embeddings used to compute the ubiquitous attention mechanisms for different neural networks architectures. By only changing the geometry of embedding of object representations, we can use the embedding space more efficiently without increasing the number of parameters of the model. Mainly as the number of objects grows exponentially for any semantic distance from the query, hyperbolic geometry --as opposed to Euclidean geometry-- can encode those objects without having any interference. Our method shows improvements in generalization on neural machine translation on WMT'14 (English to German), learning on graphs (both on synthetic and real-world graph tasks) and visual question answering (CLEVR) tasks while keeping the neural representations compact.", "target": ["Proponiamo di incorporare bias induttivi e operazioni provenienti dalla geometria iperbolica per migliorare il meccanismo di attention delle reti neurali.", "Questo articolo sostituisce la similitudine tramite prodotto scalare usata nei meccanismi di attention con la distanza iperbolica negativa, e la applica al modello di Transformer esistente, alle graph attention network e alle Relation Network", "Gli autori propongono un nuovo approccio per migliorare l'attention relazionale cambiando le funzioni di corrispondenza e aggregazione per utilizzare la geometrica iperbolica."]} +{"source": "We present a method for evaluating the sensitivity of deep reinforcement learning (RL) policies. We also formulate a zero-sum dynamic game for designing robust deep reinforcement learning policies. Our approach mitigates the brittleness of policies when agents are trained in a simulated environment and are later exposed to the real world where it is hazardous to employ RL policies. This framework for training deep RL policies involve a zero-sum dynamic game against an adversarial agent, where the goal is to drive the system dynamics to a saddle region. Using a variant of the guided policy search algorithm, our agent learns to adopt robust policies that require less samples for learning the dynamics and performs better than the GPS algorithm. Without loss of generality, we demonstrate that deep RL policies trained in this fashion will be maximally robust to a ``worst\" possible adversarial disturbances.", "target": ["Questo articolo dimostra come la teoria del controllo H-infinity può aiutare a progettare meglio policy profonde e robuste per i task di motor robot", "Propone di incorporare elementi di controllo robusto nella ricerca guidata di una policy per concepire un metodo che sia resiliente alle perturbazioni e al mismatch del modello.", "L'articolo presenta un metodo per valutare la sensibilità e la robustezza delle policy di deep RL, e propone un approccio di gioco dinamico per l'apprendimento di policy robuste."]} +{"source": "Deep networks have recently been shown to be vulnerable to universal perturbations: there exist very small image-agnostic perturbations that cause most natural images to be misclassified by such classifiers. In this paper, we provide a quantitative analysis of the robustness of classifiers to universal perturbations, and draw a formal link between the robustness to universal perturbations, and the geometry of the decision boundary. Specifically, we establish theoretical bounds on the robustness of classifiers under two decision boundary models (flat and curved models). We show in particular that the robustness of deep networks to universal perturbations is driven by a key property of their curvature: there exist shared directions along which the decision boundary of deep networks is systematically positively curved. Under such conditions, we prove the existence of small universal perturbations. Our analysis further provides a novel geometric method for computing universal perturbations, in addition to explaining their properties.", "target": ["Analisi della vulnerabilità dei classificatori alle perturbazioni universali e la relazione con la curvatura del decision boundary.", "L'articolo fornisce un'interessante analisi che collega la geometria dei confini decisionali del classificatore a piccole perturbazioni universali adversarial.", "Questo articolo discute le perturbazioni universali - perturbazioni che possono fuorviare un classificatore addestrato se aggiunte alla maggior parte dei datapoint di input.", "L'articolo sviluppa modelli che tentano di spiegare l'esistenza di perturbazioni universali che ingannano le reti neurali"]} +{"source": "Behavioral skills or policies for autonomous agents are conventionally learned from reward functions, via reinforcement learning, or from demonstrations, via imitation learning. However, both modes of task specification have their disadvantages: reward functions require manual engineering, while demonstrations require a human expert to be able to actually perform the task in order to generate the demonstration. Instruction following from natural language instructions provides an appealing alternative: in the same way that we can specify goals to other humans simply by speaking or writing, we would like to be able to specify tasks for our machines. However, a single instruction may be insufficient to fully communicate our intent or, even if it is, may be insufficient for an autonomous agent to actually understand how to perform the desired task. In this work, we propose an interactive formulation of the task specification problem, where iterative language corrections are provided to an autonomous agent, guiding it in acquiring the desired skill. Our proposed language-guided policy learning algorithm can integrate an instruction and a sequence of corrections to acquire new skills very quickly. In our experiments, we show that this method can enable a policy to follow instructions and corrections for simulated navigation and manipulation tasks, substantially outperforming direct, non-interactive instruction following.", "target": ["Proponiamo un metodo di meta-learning per la correzione interattiva delle policy con il linguaggio naturale.", "Questo articolo fornisce un framework di meta learning che mostra come apprendere nuovi task in una configurazione interattiva. Ogni task viene appreso attraverso un setup di reinforcement learning, e poi il task viene aggiornato osservando nuove istruzioni.", "Questo paper insegna agli agenti a completare task tramite istruzioni in linguaggio naturale in un processo iterativo."]} +{"source": "Deep generative models such as Generative Adversarial Networks (GANs) and Variational Auto-Encoders (VAEs) are important tools to capture and investigate the properties of complex empirical data. However, the complexity of their inner elements makes their functionment challenging to assess and modify. In this respect, these architectures behave as black box models. In order to better understand the function of such networks, we analyze their modularity based on the counterfactual manipulation of their internal variables. Our experiments on the generation of human faces with VAEs and GANs support that modularity between activation maps distributed over channels of generator architectures is achieved to some degree, can be used to better understand how these systems operate and allow meaningful transformations of the generated images without further training. erate and edit the content of generated images.", "target": ["Studiamo la modularità dei deep generative model.", "L'articolo fornisce un modo per indagare la struttura modulare dei deep generative model, con il concetto chiave di distribuire su canali di architetture generatrici."]} +{"source": "Relational databases store a significant amount of the worlds data. However, accessing this data currently requires users to understand a query language such as SQL. We propose Seq2SQL, a deep neural network for translating natural language questions to corresponding SQL queries. Our model uses rewards from in the loop query execution over the database to learn a policy to generate the query, which contains unordered parts that are less suitable for optimization via cross entropy loss. Moreover, Seq2SQL leverages the structure of SQL to prune the space of generated queries and significantly simplify the generation problem. In addition to the model, we release WikiSQL, a dataset of 80654 hand-annotated examples of questions and SQL queries distributed across 24241 tables fromWikipedia that is an order of magnitude larger than comparable datasets. By applying policy based reinforcement learning with a query execution environment to WikiSQL, Seq2SQL outperforms a state-of-the-art semantic parser, improving execution accuracy from 35.9% to 59.4% and logical form accuracy from 23.4% to 48.3%.", "target": ["Introduciamo Seq2SQL, che traduce le domande in query SQL utilizzando le reward dell'esecuzione delle query online, e WikiSQL, un dataset di tabelle/question/query SQL di diversi ordini di grandezza maggiore dei dataset esistenti.", "Un nuovo dataset di parsing semantico che si concentra sulla generazione di SQL dal linguaggio naturale utilizzando un modello basato sul reinforcement learning"]} +{"source": "We introduce Explainable Adversarial Learning, ExL, an approach for training neural networks that are intrinsically robust to adversarial attacks. We find that the implicit generative modeling of random noise with the same loss function used during posterior maximization, improves a model's understanding of the data manifold furthering adversarial robustness. We prove our approach's efficacy and provide a simplistic visualization tool for understanding adversarial data, using Principal Component Analysis. Our analysis reveals that adversarial robustness, in general, manifests in models with higher variance along the high-ranked principal components. We show that models learnt with our approach perform remarkably well against a wide-range of attacks. Furthermore, combining ExL with state-of-the-art adversarial training extends the robustness of a model, even beyond what it is adversarially trained for, in both white-box and black-box attack scenarios.", "target": ["La modellazione del rumore in ingresso durante il training discriminativo migliora l'adversarial robustness. Si propone una metrica di valutazione basata sulla PCA per l'adversarial robustness", "Questo articolo propone ExL, un metodo di adversarial training che utilizza il rumore moltiplicativo e che si dimostra utile per difendersi dagli attacchi blackbox su tre dataset.", "Questo paper include il rumore moltiplicativo N nei dati di training per ottenere l'adversarial robustness, quando si allena sia sui parametri del modello theta che sul rumore stesso."]} +{"source": "We propose a method which can visually explain the classification decision of deep neural networks (DNNs). There are many proposed methods in machine learning and computer vision seeking to clarify the decision of machine learning black boxes, specifically DNNs. All of these methods try to gain insight into why the network \"chose class A\" as an answer. Humans, when searching for explanations, ask two types of questions. The first question is, \"Why did you choose this answer? \" The second question asks, \"Why did you not choose answer B over A?\" The previously proposed methods are either not able to provide the latter directly or efficiently. We introduce a method capable of answering the second question both directly and efficiently. In this work, we limit the inputs to be images. In general, the proposed method generates explanations in the input space of any model capable of efficient evaluation and gradient evaluation. We provide results, showing the superiority of this approach for gaining insight into the inner representation of machine learning models.", "target": ["Un metodo per rispondere alla domanda \"perché non la classe B?\" per spiegare le deep network", "L'articolo propone un approccio per fornire spiegazioni visive contrastanti per le deep neural network."]} +{"source": "We flip the usual approach to study invariance and robustness of neural networks by considering the non-uniqueness and instability of the inverse mapping. We provide theoretical and numerical results on the inverse of ReLU-layers. First, we derive a necessary and sufficient condition on the existence of invariance that provides a geometric interpretation. Next, we move to robustness via analyzing local effects on the inverse. To conclude, we show how this reverse point of view not only provides insights into key effects, but also enables to view adversarial examples from different perspectives.", "target": ["Analizziamo l'invertibilità delle deep neural network studiando le preimmagini dei layer ReLU e la stabilità dell'inverso.", "Questo articolo studia il volume di preimmagine dell'attivazione di una rete ReLU ad un certo layer, e si basa sulla linearità a tratti della funzione forward di una rete ReLU.", "Questo articolo presenta un'analisi dell'invarianza inversa delle reti ReLU e fornisce limiti superiori sui valori singolari di una train network."]} +{"source": "While deep learning has led to remarkable results on a number of challenging problems, researchers have discovered a vulnerability of neural networks in adversarial settings, where small but carefully chosen perturbations to the input can make the models produce extremely inaccurate outputs. This makes these models particularly unsuitable for safety-critical application domains (e.g. self-driving cars) where robustness is extremely important. Recent work has shown that augmenting training with adversarially generated data provides some degree of robustness against test-time attacks. In this paper we investigate how this approach scales as we increase the computational budget given to the defender. We show that increasing the number of parameters in adversarially-trained models increases their robustness, and in particular that ensembling smaller models while adversarially training the entire ensemble as a single model is a more efficient way of spending said budget than simply using a larger single model. Crucially, we show that it is the adversarial training of the ensemble, rather than the ensembling of adversarially trained models, which provides robustness.", "target": ["L'adversarial training degli ensemble fornisce robustezza agli adversarial example superiore a quella osservata nei modelli addestrati in modo adversarial e negli ensemble addestrati in modo indipendente.", "Propone di addestrare congiuntamente un ensemble di modelli, dove, ad ogni passo temporale, un insieme di esempi che sono adversarial all'ensemble stesso viene incorporato nell'apprendimento."]} +{"source": "Multi-task learning (MTL) with neural networks leverages commonalities in tasks to improve performance, but often suffers from task interference which reduces the benefits of transfer. To address this issue we introduce the routing network paradigm, a novel neural network and training algorithm. A routing network is a kind of self-organizing neural network consisting of two components: a router and a set of one or more function blocks. A function block may be any neural network – for example a fully-connected or a convolutional layer. Given an input the router makes a routing decision, choosing a function block to apply and passing the output back to the router recursively, terminating when a fixed recursion depth is reached. In this way the routing network dynamically composes different function blocks for each input. We employ a collaborative multi-agent reinforcement learning (MARL) approach to jointly train the router and function blocks. We evaluate our model against cross-stitch networks and shared-layer baselines on multi-task settings of the MNIST, mini-imagenet, and CIFAR-100 datasets. Our experiments demonstrate a significant improvement in accuracy, with sharper convergence. In addition, routing networks have nearly constant per-task training cost while cross-stitch networks scale linearly with the number of tasks. On CIFAR100 (20 tasks) we obtain cross-stitch performance levels with an 85% average reduction in training time.", "target": ["Routing network: un nuovo tipo di rete neurale che impara a instradare in modo adattivo il suo input per multi-task learning", "L'articolo suggerisce di usare una rete modulare con un controllore che prende decisioni, ad ogni passo temporale, riguardo al prossimo modulo da applicare.", "L'articolo presenta una nuova formulazione per l'apprendimento dell'architettura ottimale di una rete neurale in un framework di multi-task learning utilizzando reinforcement learning multi-agente per trovare una policy e mostra un miglioramento rispetto alle architetture hard-coded con layer condivisi."]} +{"source": "We propose a practical method for $L_0$ norm regularization for neural networks: pruning the network during training by encouraging weights to become exactly zero. Such regularization is interesting since (1) it can greatly speed up training and inference, and (2) it can improve generalization. AIC and BIC, well-known model selection criteria, are special cases of $L_0$ regularization. However, since the $L_0$ norm of weights is non-differentiable, we cannot incorporate it directly as a regularization term in the objective function. We propose a solution through the inclusion of a collection of non-negative stochastic gates, which collectively determine which weights to set to zero. We show that, somewhat surprisingly, for certain distributions over the gates, the expected $L_0$ regularized objective is differentiable with respect to the distribution parameters. We further propose the \\emph{hard concrete} distribution for the gates, which is obtained by ``stretching'' a binary concrete distribution and then transforming its samples with a hard-sigmoid. The parameters of the distribution over the gates can then be jointly optimized with the original network parameters. As a result our method allows for straightforward and efficient learning of model structures with stochastic gradient descent and allows for conditional computation in a principled way. We perform various experiments to demonstrate the effectiveness of the resulting approach and regularizer.", "target": ["Mostriamo come ottimizzare la norma L_0 attesa dei modelli parametrici con la gradient descent e introduciamo una nuova distribuzione che facilita l'hard gating.", "Gli autori introducono un approccio basato sul gradiente per minimizzare una funzione obiettivo con una penalità L0 sparsa per aiutare l'apprendimento di reti neurali sparse"]} +{"source": "Recently popularized graph neural networks achieve the state-of-the-art accuracy on a number of standard benchmark datasets for graph-based semi-supervised learning, improving significantly over existing approaches. These architectures alternate between a propagation layer that aggregates the hidden states of the local neighborhood and a fully-connected layer. Perhaps surprisingly, we show that a linear model, that removes all the intermediate fully-connected layers, is still able to achieve a performance comparable to the state-of-the-art models. This significantly reduces the number of parameters, which is critical for semi-supervised learning where number of labeled examples are small. This in turn allows a room for designing more innovative propagation layers. Based on this insight, we propose a novel graph neural network that removes all the intermediate fully-connected layers, and replaces the propagation layers with attention mechanisms that respect the structure of the graph. The attention mechanism allows us to learn a dynamic and adaptive local summary of the neighborhood to achieve more accurate predictions. In a number of experiments on benchmark citation networks datasets, we demonstrate that our approach outperforms competing methods. By examining the attention weights among neighbors, we show that our model provides some interesting insights on how neighbors influence each other.", "target": ["Proponiamo una nuova architettura di Graph Neural Network interpretabile basata sull'attention che supera le attuali Graph Neural Network allo stato dell'arte in dataset standard di riferimento", "Gli autori propongono due estensioni delle GCN, rimuovendo le non linearità intermedie dal calcolo nelle GCN e aggiungendo un meccanismo di attention nel livello di aggregazione.", "L'articolo propone un algoritmo di apprendimento semi-supervised per la classificazione dei nodi dei grafi, ispirato alle Reti Neurali Grafiche."]} +{"source": "Modern generative models are usually designed to match target distributions directly in the data space, where the intrinsic dimensionality of data can be much lower than the ambient dimensionality. We argue that this discrepancy may contribute to the difficulties in training generative models. We therefore propose to map both the generated and target distributions to the latent space using the encoder of a standard autoencoder, and train the generator (or decoder) to match the target distribution in the latent space. The resulting method, perceptual generative autoencoder (PGA), is then incorporated with maximum likelihood or variational autoencoder (VAE) objective to train the generative model. With maximum likelihood, PGA generalizes the idea of reversible generative models to unrestricted neural network architectures and arbitrary latent dimensionalities. When combined with VAE, PGA can generate sharper samples than vanilla VAE.", "target": ["Un framework per il training di modelli generativi basati su autoencoder, con loss non-adversarial e architetture di reti neurali non limitate.", "Questo articolo usa autoencoder per fare il matching di distribuzioni in uno spazio ad alta dimensione."]} +{"source": "The quality of the representations achieved by embeddings is determined by how well the geometry of the embedding space matches the structure of the data. Euclidean space has been the workhorse for embeddings; recently hyperbolic and spherical spaces have gained popularity due to their ability to better embed new types of structured data---such as hierarchical data---but most data is not structured so uniformly. We address this problem by proposing learning embeddings in a product manifold combining multiple copies of these model spaces (spherical, hyperbolic, Euclidean), providing a space of heterogeneous curvature suitable for a wide variety of structures. We introduce a heuristic to estimate the sectional curvature of graph data and directly determine an appropriate signature---the number of component spaces and their dimensions---of the product manifold. Empirically, we jointly learn the curvature and the embedding in the product space via Riemannian optimization. We discuss how to define and compute intrinsic quantities such as means---a challenging notion for product manifolds---and provably learnable optimization functions. On a range of datasets and reconstruction tasks, our product space embeddings outperform single Euclidean or hyperbolic spaces used in previous works, reducing distortion by 32.55% on a Facebook social network dataset. We learn word embeddings and find that a product of hyperbolic spaces in 50 dimensions consistently improves on baseline Euclidean and hyperbolic embeddings, by 2.6 points in Spearman rank correlation on similarity tasks and 3.4 points on analogy accuracy.", "target": ["Gli spazi di embedding di product manifold con curvatura eterogenea producono rappresentazioni migliori rispetto agli spazi di embedding tradizionali per una varietà di strutture.", "Propone un metodo di riduzione della dimensionalità che incorpora i dati in un product manifold di manifold sferici, euclidei e iperbolici. L'algoritmo si basa sul matching tra le distanze geodetiche sul product manifold e le distanze nei grafi."]} +{"source": "Synthesizing user-intended programs from a small number of input-output exam- ples is a challenging problem with several important applications like spreadsheet manipulation, data wrangling and code refactoring. Existing synthesis systems either completely rely on deductive logic techniques that are extensively hand- engineered or on purely statistical models that need massive amounts of data, and in general fail to provide real-time synthesis on challenging benchmarks. In this work, we propose Neural Guided Deductive Search (NGDS), a hybrid synthesis technique that combines the best of both symbolic logic techniques and statistical models. Thus, it produces programs that satisfy the provided specifications by construction and generalize well on unseen examples, similar to data-driven systems. Our technique effectively utilizes the deductive search framework to reduce the learning problem of the neural component to a simple supervised learning setup. Further, this allows us to both train on sparingly available real-world data and still leverage powerful recurrent neural network encoders. We demonstrate the effectiveness of our method by evaluating on real-world customer scenarios by synthesizing accurate programs with up to 12× speed-up compared to state-of-the-art systems.", "target": ["Integriamo metodi simbolici (deduttivi) e statistici (neural-based) per permettere la sintesi di programmi in tempo reale con una generalizzazione quasi perfetta da 1 esempio di input-output.", "L'articolo presenta un approccio branch-and-bound per imparare buoni programmi dove una LSTM è usato per predire quali rami nell'albero di ricerca dovrebbero portare a buoni programmi", "Propone un sistema che sintetizza programmi da un singolo esempio che generalizza meglio dello stato dell'arte precedente"]} +{"source": "Variational auto-encoders (VAEs) offer a tractable approach when performing approximate inference in otherwise intractable generative models. However, standard VAEs often produce latent codes that are disperse and lack interpretability, thus making the resulting representations unsuitable for auxiliary tasks (e.g. classification) and human interpretation. We address these issues by merging ideas from variational auto-encoders and sparse coding, and propose to explicitly model sparsity in the latent space of a VAE with a Spike and Slab prior distribution. We derive the evidence lower bound using a discrete mixture recognition function thereby making approximate posterior inference as computational efficient as in the standard VAE case. With the new approach, we are able to infer truly sparse representations with generally intractable non-linear probabilistic models. We show that these sparse representations are advantageous over standard VAE representations on two benchmark classification tasks (MNIST and Fashion-MNIST) by demonstrating improved classification accuracy and significantly increased robustness to the number of latent dimensions. Furthermore, we demonstrate qualitatively that the sparse elements capture subjectively understandable sources of variation.", "target": ["Esploriamo l'intersezione di VAE e sparse coding.", "Questo articolo propone un'estensione dei VAE con prior e posterior sparsi per imparare rappresentazioni sparse interpretabili."]} +{"source": "A widely observed phenomenon in deep learning is the degradation problem: increasing the depth of a network leads to a decrease in performance on both test and training data. Novel architectures such as ResNets and Highway networks have addressed this issue by introducing various flavors of skip-connections or gating mechanisms. However, the degradation problem persists in the context of plain feed-forward networks. In this work we propose a simple method to address this issue. The proposed method poses the learning of weights in deep networks as a constrained optimization problem where the presence of skip-connections is penalized by Lagrange multipliers. This allows for skip-connections to be introduced during the early stages of training and subsequently phased out in a principled manner. We demonstrate the benefits of such an approach with experiments on MNIST, fashion-MNIST, CIFAR-10 and CIFAR-100 where the proposed method is shown to greatly decrease the degradation effect (compared to plain networks) and is often competitive with ResNets.", "target": ["L'eliminazione graduale delle skip-connection evita la degradazione nelle deep feed-forward network.", "Gli autori presentano una nuova strategia di training, VAN, per il training di reti feed-forward molto profonde senza skip-connection", "L'articolo introduce un'architettura che interpola linearmente tra ResNet e le deep network standard senza skip-connection."]} +{"source": "Deep learning is becoming more widespread in its application due to its power in solving complex classification problems. However, deep learning models often require large memory and energy consumption, which may prevent them from being deployed effectively on embedded platforms, limiting their applications. This work addresses the problem by proposing methods {\\em Weight Reduction Quantisation} for compressing the memory footprint of the models, including reducing the number of weights and the number of bits to store each weight. Beside, applying with sparsity-inducing regularization, our work focuses on speeding up stochastic variance reduced gradients (SVRG) optimization on non-convex problem. Our method that mini-batch SVRG with $\\ell$1 regularization on non-convex problem has faster and smoother convergence rates than SGD by using adaptive learning rates. Experimental evaluation of our approach uses MNIST and CIFAR-10 datasets on LeNet-300-100 and LeNet-5 models, showing our approach can reduce the memory requirements both in the convolutional and fully connected layers by up to 60$\\times$ without affecting their test accuracy.", "target": ["Compressione di deep neural network distribuite su dispositivi embedded.", "Gli autori presentano un algoritmo di training basato su SVRG regolarizzato l-1 che è in grado di forzare molti pesi della rete ad essere pari a 0.", "Questo lavoro riduce i requisiti di memoria."]} +{"source": "It has been argued that the brain is a prediction machine that continuously learns how to make better predictions about the stimuli received from the external environment. For this purpose, it builds a model of the world around us and uses this model to infer the external stimulus. Predictive coding has been proposed as a mechanism through which the brain might be able to build such a model of the external environment. However, it is not clear how predictive coding can be used to build deep neural network models of the brain while complying with the architectural constraints imposed by the brain. In this paper, we describe an algorithm to build a deep generative model using predictive coding that can be used to infer latent representations about the stimuli received from external environment. Specifically, we used predictive coding to train a deep neural network on real-world images in a unsupervised learning paradigm. To understand the capacity of the network with regards to modeling the external environment, we studied the latent representations generated by the model on images of objects that are never presented to the model during training. Despite the novel features of these objects the model is able to infer the latent representations for them. Furthermore, the reconstructions of the original images obtained from these latent representations preserve the important details of these objects.", "target": ["Un algoritmo di apprendimento basato sulla codifica predittiva per costruire modelli di deep neural network del cervello", "L'articolo considera l'apprendimento di una rete neurale generativa utilizzando una configurazione di codifica predittiva"]} +{"source": "In this paper, we propose deep convolutional generative adversarial networks (DCGAN) that learn to produce a 'mental image' of the input image as internal representation of a certain category of input data distribution. This mental image is what the DCGAN 'imagines' that the input image might look like under ideal conditions. The mental image contains a version of the input that is iconic, without any peculiarities that do not contribute to the ideal representation of the input data distribution within a category. A DCGAN learns this association by training an encoder to capture salient features from the original image and a decoder to convert salient features into its associated mental image representation. Our new approach, which we refer to as a Mental Image DCGAN (MIDCGAN), learns features that are useful for recognizing entire classes of objects, and that this in turn has the benefit of helping single and zero shot recognition. We demonstrate our approach on object instance recognition and handwritten digit recognition tasks.", "target": ["Il riconoscimento di istanze di oggetti con adversarial autoencoder è stato eseguito con un nuovo obiettivo di \"mental image\" che è una rappresentazione canonica dell'immagine di input.", "L'articolo propone un metodo per imparare le feature per il riconoscimento degli oggetti che è invariante a varie trasformazioni dell'oggetto, in particolare alla posa dell'oggetto.", "Questo articolo ha studiato il task di riconoscimento few shot tramite un'immagine mentale generata come rappresentazione intermedia data l'immagine di input."]} +{"source": "An obstacle that prevents the wide adoption of (deep) reinforcement learning (RL) in control systems is its need for a large number of interactions with the environment in order to master a skill. The learned skill usually generalizes poorly across domains and re-training is often necessary when presented with a new task. We present a framework that combines techniques in \\textit{formal methods} with \\textit{hierarchical reinforcement learning} (HRL). The set of techniques we provide allows for the convenient specification of tasks with logical expressions, learns hierarchical policies (meta-controller and low-level controllers) with well-defined intrinsic rewards using any RL methods and is able to construct new skills from existing ones without additional learning. We evaluate the proposed methods in a simple grid world simulation as well as simulation on a Baxter robot.", "target": ["Si combina la logica temporale con il reinforcement learning gerarchico per la composizione delle skill", "L'articolo offre una strategia per costruire un MDP prodotto da un MDP originale e dall'automa associato a una formula LTL.", "Propone di unire la logica temporale con il reinforcement learning gerarchico per semplificare la composizione delle skill."]} +{"source": "The tremendous memory and computational complexity of Convolutional Neural Networks (CNNs) prevents the inference deployment on resource-constrained systems. As a result, recent research focused on CNN optimization techniques, in particular quantization, which allows weights and activations of layers to be represented with just a few bits while achieving impressive prediction performance. However, aggressive quantization techniques still fail to achieve full-precision prediction performance on state-of-the-art CNN architectures on large-scale classification tasks. In this work we propose a method for weight and activation quantization that is scalable in terms of quantization levels (n-ary representations) and easy to compute while maintaining the performance close to full-precision CNNs. Our weight quantization scheme is based on trainable scaling factors and a nested-means clustering strategy which is robust to weight updates and therefore exhibits good convergence properties. The flexibility of nested-means clustering enables exploration of various n-ary weight representations with the potential of high parameter compression. For activations, we propose a linear quantization strategy that takes the statistical properties of batch normalization into account. We demonstrate the effectiveness of our approach using state-of-the-art models on ImageNet.", "target": ["Proponiamo uno schema di quantizzazione per i pesi e le attivazioni delle deep neural network. Questo riduce l'impronta di memoria in modo sostanziale e accelera l'inferenza.", "Compressione del modello CNN e accelerazione dell'inferenza usando la quantizzazione."]} +{"source": "Reinforcement learning (RL) agents optimize only the features specified in a reward function and are indifferent to anything left out inadvertently. This means that we must not only specify what to do, but also the much larger space of what not to do. It is easy to forget these preferences, since these preferences are already satisfied in our environment. This motivates our key insight: when a robot is deployed in an environment that humans act in, the state of the environment is already optimized for what humans want. We can therefore use this implicit preference information from the state to fill in the blanks. We develop an algorithm based on Maximum Causal Entropy IRL and use it to evaluate the idea in a suite of proof-of-concept environments designed to show its properties. We find that information from the initial state can be used to infer both side effects that should be avoided as well as preferences for how the environment should be organized. Our code can be found at https://github.com/HumanCompatibleAI/rlsp.", "target": ["Quando un robot viene schierato in un ambiente in cui gli umani hanno agito, lo stato dell'ambiente è già ottimizzato per ciò che gli umani vogliono e possiamo usare questo fatto per dedurre le preferenze umane.", "Gli autori propongono di arricchire la funzione di reward esplicitamente dichiarata di un agente RL con reward/costi ausiliari dedotti dallo stato iniziale e un modello della dinamica dello stato", "Questo lavoro propone un modo per dedurre le informazioni implicite nello stato iniziale usando IRL e per combinare la reward dedotta con una reward specificata."]} +{"source": "Regularization is one of the crucial ingredients of deep learning, yet the term regularization has various definitions, and regularization methods are often studied separately from each other. In our work we present a novel, systematic, unifying taxonomy to categorize existing methods. We distinguish methods that affect data, network architectures, error terms, regularization terms, and optimization procedures. We identify the atomic building blocks of existing methods, and decouple the assumptions they enforce from the mathematical tools they rely on. We do not provide all details about the listed methods; instead, we present an overview of how the methods can be sorted into meaningful categories and sub-categories. This helps revealing links and fundamental similarities between them. Finally, we include practical recommendations both for users and for developers of new regularization methods.", "target": ["Categorizzazione sistematica dei metodi di regolarizzazione per il deep learning, rivelando le loro somiglianze.", "Tenta di costruire una tassonomia per le tecniche di regolarizzazione impiegate nel deep learning."]} +{"source": "Deep neural networks are surprisingly efficient at solving practical tasks, but the theory behind this phenomenon is only starting to catch up with the practice. Numerous works show that depth is the key to this efficiency. A certain class of deep convolutional networks – namely those that correspond to the Hierarchical Tucker (HT) tensor decomposition – has been proven to have exponentially higher expressive power than shallow networks. I.e. a shallow network of exponential width is required to realize the same score function as computed by the deep architecture. In this paper, we prove the expressive power theorem (an exponential lower bound on the width of the equivalent shallow network) for a class of recurrent neural networks – ones that correspond to the Tensor Train (TT) decomposition. This means that even processing an image patch by patch with an RNN can be exponentially more efficient than a (shallow) convolutional network with one hidden layer. Using theoretical results on the relation between the tensor decompositions we compare expressive powers of the HT- and TT-Networks. We also implement the recurrent TT-Networks and provide numerical evidence of their expressivity.", "target": ["Dimostriamo l'efficienza esponenziale delle reti neurali di tipo ricorrente rispetto alle reti poco profonde.", "Gli autori confrontano la complessità delle reti tensor train con reti strutturate dalla decomposizione CP"]} +{"source": "Probabilistic modelling is a principled framework to perform model aggregation, which has been a primary mechanism to combat mode collapse in the context of Generative Adversarial Networks (GAN). In this paper, we propose a novel probabilistic framework for GANs, ProbGAN, which iteratively learns a distribution over generators with a carefully crafted prior. Learning is efficiently triggered by a tailored stochastic gradient Hamiltonian Monte Carlo with a novel gradient approximation to perform Bayesian inference. Our theoretical analysis further reveals that our treatment is the first probabilistic framework that yields an equilibrium where generator distributions are faithful to the data distribution. Empirical evidence on synthetic high-dimensional multi-modal data and image databases (CIFAR-10, STL-10, and ImageNet) demonstrates the superiority of our method over both start-of-the-art multi-generator GANs and other probabilistic treatment for GANs.", "target": ["Un nuovo trattamento probabilistico per le GAN con garanzie teoriche.", "Questo articolo propone una GAN bayesiana che ha garanzie teoriche di convergenza alla distribuzione reale e pone le verosimiglianze sul generatore e sul discriminatore con logaritmi proporzionali alle funzioni obiettivo delle GAN tradizionali."]} +{"source": "In the adversarial-perturbation problem of neural networks, an adversary starts with a neural network model $F$ and a point $\\bfx$ that $F$ classifies correctly, and applies a \\emph{small perturbation} to $\\bfx$ to produce another point $\\bfx'$ that $F$ classifies \\emph{incorrectly}. In this paper, we propose taking into account \\emph{the inherent confidence information} produced by models when studying adversarial perturbations, where a natural measure of ``confidence'' is \\|F(\\bfx)\\|_\\infty$ (i.e. how confident $F$ is about its prediction?) . Motivated by a thought experiment based on the manifold assumption, we propose a ``goodness property'' of models which states that \\emph{confident regions of a good model should be well separated}. We give formalizations of this property and examine existing robust training objectives in view of them. Interestingly, we find that a recent objective by Madry et al. encourages training a model that satisfies well our formal version of the goodness property, but has a weak control of points that are wrong but with low confidence. However, if Madry et al.'s model is indeed a good solution to their objective, then good and bad points are now distinguishable and we can try to embed uncertain points back to the closest confident region to get (hopefully) correct predictions. We thus propose embedding objectives and algorithms, and perform an empirical study using this method. Our experimental results are encouraging: Madry et al.'s model wrapped with our embedding procedure achieves almost perfect success rate in defending against attacks that the base model fails on, while retaining good generalization behavior.", "target": ["Difesa contro le adversarial perturbation delle reti neurali a partire dall'assunzione del manifold", "Il manoscritto propone due funzioni obiettivo basate sull'assunzione del manifold come meccanismi di difesa contro gli adversarial example.", "Difendersi dagli adversarial attack basati sull'assunzione del manifold dei dati naturali"]} +{"source": "Recently Neural Architecture Search (NAS) has aroused great interest in both academia and industry, however it remains challenging because of its huge and non-continuous search space. Instead of applying evolutionary algorithm or reinforcement learning as previous works, this paper proposes a Direct Sparse Optimization NAS (DSO-NAS) method. In DSO-NAS, we provide a novel model pruning view to NAS problem. In specific, we start from a completely connected block, and then introduce scaling factors to scale the information flow between operations. Next, we impose sparse regularizations to prune useless connections in the architecture. Lastly, we derive an efficient and theoretically sound optimization method to solve it. Our method enjoys both advantages of differentiability and efficiency, therefore can be directly applied to large datasets like ImageNet. Particularly, On CIFAR-10 dataset, DSO-NAS achieves an average test error 2.84%, while on the ImageNet dataset DSO-NAS achieves 25.4% test error under 600M FLOPs with 8 GPUs in 18 hours.", "target": ["Ricerca dell'architettura neurale single shot tramite ottimizzazione diretta sparsa", "Presenta un metodo di ricerca dell'architettura in cui le connessioni vengono rimosse con una regolarizzazione sparsa.", "Questo articolo propone Direct Sparse Optimization, che è un metodo per ottenere architetture neurali su problemi specifici ad un costo computazionale ragionevole.", "Questo articolo propone un metodo di ricerca dell'architettura neurale basato su un'ottimizzazione sparsa diretta"]} +{"source": "Deep neural networks (DNNs) continue to make significant advances, solving tasks from image classification to translation or reinforcement learning. One aspect of the field receiving considerable attention is efficiently executing deep models in resource-constrained environments, such as mobile or embedded devices. This paper focuses on this problem, and proposes two new compression methods, which jointly leverage weight quantization and distillation of larger teacher networks into smaller student networks. The first method we propose is called quantized distillation and leverages distillation during the training process, by incorporating distillation loss, expressed with respect to the teacher, into the training of a student network whose weights are quantized to a limited set of levels. The second method, differentiable quantization, optimizes the location of quantization points through stochastic gradient descent, to better fit the behavior of the teacher model. We validate both methods through experiments on convolutional and recurrent architectures. We show that quantized shallow students can reach similar accuracy levels to full-precision teacher models, while providing order of magnitude compression, and inference speedup that is linear in the depth reduction. In sum, our results enable DNNs for resource-constrained environments to leverage architecture and accuracy advances developed on more powerful devices.", "target": ["Ottiene una precisione allo stato dell'arte per reti quantizzate e poco profonde sfruttando la distillation.", "Propone modelli piccoli e a basso costo combinando distillation e quantizzazione per esperimenti di computer vision e neural machine translation", "Questo articolo presenta un framework di utilizzo di un teacher model per aiutare la compressione del modello di deep learning nel contesto della compressione del modello."]} +{"source": "Previous work has demonstrated the benefits of incorporating additional linguistic annotations such as syntactic trees into neural machine translation. However the cost of obtaining those syntactic annotations is expensive for many languages and the quality of unsupervised learning linguistic structures is too poor to be helpful. In this work, we aim to improve neural machine translation via source side dependency syntax but without explicit annotation. We propose a set of models that learn to induce dependency trees on the source side and learn to use that information on the target side. Importantly, we also show that our dependency trees capture important syntactic features of language and improve translation quality on two language pairs En-De and En-Ru.", "target": ["Migliorare la NMT con gli alberi latenti", "Questo articolo descrive un metodo per indurre strutture di dipendenza source-side al servizio della traduzione automatica neurale."]} +{"source": "Model-free reinforcement learning (RL) requires a large number of trials to learn a good policy, especially in environments with sparse rewards. We explore a method to improve the sample efficiency when we have access to demonstrations. Our approach, Backplay, uses a single demonstration to construct a curriculum for a given task. Rather than starting each training episode in the environment's fixed initial state, we start the agent near the end of the demonstration and move the starting point backwards during the course of training until we reach the initial state. Our contributions are that we analytically characterize the types of environments where Backplay can improve training speed, demonstrate the effectiveness of Backplay both in large grid worlds and a complex four player zero-sum game (Pommerman), and show that Backplay compares favorably to other competitive methods known to improve sample efficiency. This includes reward shaping, behavioral cloning, and reverse curriculum generation.", "target": ["Imparare lavorando a ritroso da una singola dimostrazione, anche una inefficiente, e progressivamente delegare all'agente una parte sempre maggiore della risoluzione.", "Questo articolo presenta un metodo per aumentare l'efficienza dei metodi di RL a reward sparsa attraverso un curriculum backward sulle dimostrazioni di esperti.", "L'articolo presenta una strategia per risolvere task di reward sparsi con RL campionando gli stati iniziali dalle dimostrazioni."]} +{"source": "Episodic memory is a psychology term which refers to the ability to recall specific events from the past. We suggest one advantage of this particular type of memory is the ability to easily assign credit to a specific state when remembered information is found to be useful. Inspired by this idea, and the increasing popularity of external memory mechanisms to handle long-term dependencies in deep learning systems, we propose a novel algorithm which uses a reservoir sampling procedure to maintain an external memory consisting of a fixed number of past states. The algorithm allows a deep reinforcement learning agent to learn online to preferentially remember those states which are found to be useful to recall later on. Critically this method allows for efficient online computation of gradient estimates with respect to the write process of the external memory. Thus unlike most prior mechanisms for external memory it is feasible to use in an online reinforcement learning setting.", "target": ["Memoria esterna per reinforcement learning online basato sulla stima dei gradienti su una nuova tecnica di reservoir sampling.", "L'articolo propone un approccio modificato alla RL, dove una \"memoria episodica\" aggiuntiva è mantenuta dall'agente e usa una \"query network\" che si basa sullo stato attuale."]} +{"source": "We achieve bias-variance decomposition for Boltzmann machines using an information geometric formulation. Our decomposition leads to an interesting phenomenon that the variance does not necessarily increase when more parameters are included in Boltzmann machines, while the bias always decreases. Our result gives a theoretical evidence of the generalization ability of deep learning architectures because it provides the possibility of increasing the representation power with avoiding the variance inflation.", "target": ["Deduciamo la decomposizione bias-varianza per le macchine di Boltzmann usando una formulazione geometrica dell'informazione.", "L'obiettivo di questo articolo è analizzare l'efficacia e la generalizzabilità del deep learning presentando un'analisi teorica della decomposizione bias-varianza per i modelli gerarchici, in particolare le macchine di Boltzmann", "L'articolo arriva alla conclusione principale che è possibile ridurre sia il bias sia la varianza in un modello gerarchico."]} +{"source": "Recurrent Neural Networks (RNNs) are powerful tools for solving sequence-based problems, but their efficacy and execution time are dependent on the size of the network. Following recent work in simplifying these networks with model pruning and a novel mapping of work onto GPUs, we design an efficient implementation for sparse RNNs. We investigate several optimizations and tradeoffs: Lamport timestamps, wide memory loads, and a bank-aware weight layout. With these optimizations, we achieve speedups of over 6x over the next best algorithm for a hidden layer of size 2304, batch size of 4, and a density of 30%. Further, our technique allows for models of over 5x the size to fit on a GPU for a speedup of 2x, enabling larger networks to help advance the state-of-the-art. We perform case studies on NMT and speech recognition tasks in the appendix, accelerating their recurrent layers by up to 3x.", "target": ["Combinando il pruning della rete e i kernel persistenti in un'implementazione della rete pratica, veloce e accurata.", "Questo articolo introduce le RNN sparse e persistenti, un meccanismo per aggiungere il pruning al lavoro esistente di memorizzazione dei pesi delle RNN su un chip."]} +{"source": "Weight pruning has proven to be an effective method in reducing the model size and computation cost while not sacrificing the model accuracy. Conventional sparse matrix formats, however, involve irregular index structures with large storage requirement and sequential reconstruction process, resulting in inefficient use of highly parallel computing resources. Hence, pruning is usually restricted to inference with a batch size of one, for which an efficient parallel matrix-vector multiplication method exists. In this paper, a new class of sparse matrix representation utilizing Viterbi algorithm that has a high, and more importantly, fixed index compression ratio regardless of the pruning rate, is proposed. In this approach, numerous sparse matrix candidates are first generated by the Viterbi encoder, and then the one that aims to minimize the model accuracy degradation is selected by the Viterbi algorithm. The model pruning process based on the proposed Viterbi encoder and Viterbi algorithm is highly parallelizable, and can be implemented efficiently in hardware to achieve low-energy, high-performance index decoding process. Compared with the existing magnitude-based pruning methods, index data storage requirement can be further compressed by 85.2% in MNIST and 83.9% in AlexNet while achieving similar pruning rate. Even compared with the relative index compression technique, our method can still reduce the index storage requirement by 52.7% in MNIST and 35.5% in AlexNet.", "target": ["Presentiamo un nuovo metodo di pruning e un formato di matrice sparsa per permettere un alto rapporto di compressione dell'indice e un processo parallelo di decodifica dello stesso.", "Gli autori usano la codifica Viterbi per comprimere drasticamente l'indice della matrice sparsa di una rete pruned, riducendo uno dei principali sovraccarichi di memoria e accelerando l'inferenza nel setting parallelo."]} +{"source": "Learning policies for complex tasks that require multiple different skills is a major challenge in reinforcement learning (RL). It is also a requirement for its deployment in real-world scenarios. This paper proposes a novel framework for efficient multi-task reinforcement learning. Our framework trains agents to employ hierarchical policies that decide when to use a previously learned policy and when to learn a new skill. This enables agents to continually acquire new skills during different stages of training. Each learned task corresponds to a human language description. Because agents can only access previously learned skills through these descriptions, the agent can always provide a human-interpretable description of its choices. In order to help the agent learn the complex temporal dependencies necessary for the hierarchical policy, we provide it with a stochastic temporal grammar that modulates when to rely on previously learned skills and when to execute new skills. We validate our approach on Minecraft games designed to explicitly test the ability to reuse previously learned skills while simultaneously learning new skills.", "target": ["Una nuova hierarchical policy network che può riutilizzare le skill precedentemente apprese insieme e come sottocomponenti di nuove skill, scoprendo le relazioni sottostanti tra le skill.", "Questo articolo mira ad apprendere policy gerarchiche utilizzando una struttura di policy ricorsiva regolata da una grammatica temporale stocastica", "Questo articolo propone un approccio all'apprendimento delle policy gerarchiche in un contesto di lifelong learning, impilando le policy e poi usando una policy esplicita di \"switch\"."]} +{"source": "Embeddings are a fundamental component of many modern machine learning and natural language processing models. Understanding them and visualizing them is essential for gathering insights about the information they capture and the behavior of the models. State of the art in analyzing embeddings consists in projecting them in two-dimensional planes without any interpretable semantics associated to the axes of the projection, which makes detailed analyses and comparison among multiple sets of embeddings challenging. In this work, we propose to use explicit axes defined as algebraic formulae over embeddings to project them into a lower dimensional, but semantically meaningful subspace, as a simple yet effective analysis and visualization methodology. This methodology assigns an interpretable semantics to the measures of variability and the axes of visualizations, allowing for both comparisons among different sets of embeddings and fine-grained inspection of the embedding spaces. We demonstrate the power of the proposed methodology through a series of case studies that make use of visualizations constructed around the underlying methodology and through a user study. The results show how the methodology is effective at providing more profound insights than classical projection methods and how it is widely applicable to many other use cases.", "target": ["Proponiamo di usare la proiezione esplicita dalle formule algebriche vettoriali come un modo alternativo per visualizzare gli spazi di embedding che è specificamente adatto ai task di analisi goal-oriented e che supera t-SNE nel nostro user study.", "Analisi degli spazi di embedding in modo non parametrico (basato su esempi)"]} +{"source": "Learning deep networks which can resist large variations between training andtesting data is essential to build accurate and robust image classifiers. Towardsthis end, a typical strategy is to apply data augmentation to enlarge the trainingset. However, standard data augmentation is essentially a brute-force strategywhich is inefficient, as it performs all the pre-defined transformations to everytraining sample. In this paper, we propose a principled approach to train networkswith significantly improved resistance to large variations between training andtesting data. This is achieved by embedding a learnable transformation moduleinto the introspective networks (Jin et al., 2017; Lazarow et al., 2017; Lee et al.,2018), which is a convolutional neural network (CNN) classifier empowered withgenerative capabilities. Our approach alternatively synthesizes pseudo-negativesamples with learned transformations and enhances the classifier by retraining itwith synthesized samples. Experimental results verify that our approach signif-icantly improves the ability of deep networks to resist large variations betweentraining and testing data and achieves classification accuracy improvements onseveral benchmark datasets, including MNIST, affNIST, SVHN and CIFAR-10.", "target": ["Proponiamo un approccio che conferisce ai classificatori la capacità di resistere a variazioni più grandi tra i dati di training e di test in modo intelligente ed efficiente.", "Usiamo l'introspective learning per gestire le variazioni dei dati al momento del test", "Questo articolo suggerisce l'uso di transformation network apprese, incorporate in introspective network per migliorare le prestazioni di classificazione con esempi sintetizzati."]} +{"source": "It is well known that it is possible to construct \"adversarial examples\" for neural networks: inputs which are misclassified by the network yet indistinguishable from true data. We propose a simple modification to standard neural network architectures, thermometer encoding, which significantly increases the robustness of the network to adversarial examples. We demonstrate this robustness with experiments on the MNIST, CIFAR-10, CIFAR-100, and SVHN datasets, and show that models with thermometer-encoded inputs consistently have higher accuracy on adversarial examples, without decreasing generalization. State-of-the-art accuracy under the strongest known white-box attack was increased from 93.20% to 94.30% on MNIST and 50.00% to 79.16% on CIFAR-10. We explore the properties of these networks, providing evidence that thermometer encodings help neural networks to find more-non-linear decision boundaries.", "target": ["La discretizzazione dell'input porta alla robustezza contro gli adversarial example", "Gli autori presentano uno studio approfondito sulla discretizzazione/quantizzazione dell'input come difesa contro gli adversarial example"]} +{"source": "Low-precision training is a promising way of decreasing the time and energy cost of training machine learning models. Previous work has analyzed low-precision training algorithms, such as low-precision stochastic gradient descent, and derived theoretical bounds on their convergence rates. These bounds tend to depend on the dimension of the model $d$ in that the number of bits needed to achieve a particular error bound increases as $d$ increases. This is undesirable because a motivating application for low-precision training is large-scale models, such as deep learning, where $d$ can be huge. In this paper, we prove dimension-independent bounds for low-precision training algorithms that use fixed-point arithmetic, which lets us better understand what affects the convergence of these algorithms as parameters scale. Our methods also generalize naturally to let us prove new convergence bounds on low-precision training with other quantization schemes, such as low-precision floating-point computation and logarithmic quantization.", "target": ["Abbiamo dimostrato bound indipendenti dalla dimensione per algoritmi di training a bassa precisione", "Questo articolo discute le condizioni in cui la convergenza dei modelli nel training con pesi di bassa precisione non dipende dalla dimensione del modello."]} +{"source": "We consider the problem of exploration in meta reinforcement learning. Two new meta reinforcement learning algorithms are suggested: E-MAML and ERL2. Results are presented on a novel environment we call 'Krazy World' and a set of maze environments. We show E-MAML and ERL2 deliver better performance on tasks where exploration is important.", "target": ["Modifiche a MAML e RL2 che dovrebbero permettere una migliore esplorazione.", "L'articolo propone un trucco per estendere le funzioni obiettivo per guidare l'esplorazione in meta-RL in cima a due recenti algoritmi meta-RL"]} +{"source": "We propose a new class of probabilistic neural-symbolic models for visual question answering (VQA) that provide interpretable explanations of their decision making in the form of programs, given a small annotated set of human programs. The key idea of our approach is to learn a rich latent space which effectively propagates program annotations from known questions to novel questions. We do this by formalizing prior work on VQA, called module networks (Andreas, 2016) as discrete, structured, latent variable models on the joint distribution over questions and answers given images, and devise a procedure to train the model effectively. Our results on a dataset of compositional questions about SHAPES (Andreas, 2016) show that our model generates more interpretable programs and obtains better accuracy on VQA in the low-data regime than prior work.", "target": ["Un modello simbolico neurale probabilistico con uno spazio di programma latente, per una question answering più interpretabile", "Questo articolo propone un modello a variabili latenti discreto e strutturato per visual question answering che coinvolge la generalizzazione e il ragionamento compositivo con un guadagno significativo in termini di prestazioni e capacità."]} +{"source": "The ability to deploy neural networks in real-world, safety-critical systems is severely limited by the presence of adversarial examples: slightly perturbed inputs that are misclassified by the network. In recent years, several techniques have been proposed for training networks that are robust to such examples; and each time stronger attacks have been devised, demonstrating the shortcomings of existing defenses. This highlights a key difficulty in designing an effective defense: the inability to assess a network's robustness against future attacks. We propose to address this difficulty through formal verification techniques. We construct ground truths: adversarial examples with a provably-minimal distance from a given input point. We demonstrate how ground truths can serve to assess the effectiveness of attack techniques, by comparing the adversarial examples produced by those attacks to the ground truths; and also of defense techniques, by computing the distance to the ground truths before and after the defense is applied, and measuring the improvement. We use this technique to assess recently suggested attack and defense techniques.", "target": ["Usiamo la verifica formale per valutare l'efficacia delle tecniche per trovare o difendersi dagli adversarial example.", "Questo articolo propone un metodo per calcolare adversarial example con la minima distanza dagli input originali.", "Gli autori propongono di impiegare esempi a distanza minima come strumento per valutare la robustezza di una rete addestrata.", "L'articolo descrive un metodo per generare adversarial example che hanno una distanza minima dal training example usato per generarli"]} +{"source": "This paper introduces a new framework for open-domain question answering in which the retriever and the reader \\emph{iteratively interact} with each other. The framework is agnostic to the architecture of the machine reading model provided it has \\emph{access} to the token-level hidden representations of the reader. The retriever uses fast nearest neighbor search that allows it to scale to corpora containing millions of paragraphs. A gated recurrent unit updates the query at each step conditioned on the \\emph{state} of the reader and the \\emph{reformulated} query is used to re-rank the paragraphs by the retriever. We conduct analysis and show that iterative interaction helps in retrieving informative paragraphs from the corpus. Finally, we show that our multi-step-reasoning framework brings consistent improvement when applied to two widely used reader architectures (\\drqa and \\bidaf) on various large open-domain datasets ---\\tqau, \\quasart, \\searchqa, and \\squado\\footnote{Code and pretrained models are available at \\url{https://github.com/rajarshd/Multi-Step-Reasoning}}.", "target": ["Paragraph retriever e machine reader interagiscono l'uno con l'altro attraverso reinforcement learning per produrre grandi miglioramenti su dataset open domain", "L'articolo introduce un nuovo framework di interazione bidirezionale tra il paragraph retriever e il machine reader per open-domain question answering con l'idea di \"reader state\" dal reader al retriever.", "L'articolo propone un modello di multi-document extractive machine reading composto da 3 parti distinte e un algoritmo."]} +{"source": "Many imaging tasks require global information about all pixels in an image. Conventional bottom-up classification networks globalize information by decreasing resolution; features are pooled and down-sampled into a single output. But for semantic segmentation and object detection tasks, a network must provide higher-resolution pixel-level outputs. To globalize information while preserving resolution, many researchers propose the inclusion of sophisticated auxiliary blocks, but these come at the cost of a considerable increase in network size and computational cost. This paper proposes stacked u-nets (SUNets), which iteratively combine features from different resolution scales while maintaining resolution. SUNets leverage the information globalization power of u-nets in a deeper net- work architectures that is capable of handling the complexity of natural images. SUNets perform extremely well on semantic segmentation tasks using a small number of parameters.", "target": ["Presenta una nuova architettura che sfrutta il potere di globalizzazione dell'informazione delle u-net in una rete più profonda ed esegue bene tutti i task.", "Un'architettura di rete per la segmentazione semantica delle immagini, basata sulla composizione di una pila di architetture U-Net di base, che riduce il numero di parametri e migliora i risultati.", "Questo paper propone un'architettura U-Net impilata per la segmentazione delle immagini."]} +{"source": "Asking questions is an important ability for a chatbot. This paper focuses on question generation. Although there are existing works on question generation based on a piece of descriptive text, it remains to be a very challenging problem. In the paper, we propose a new question generation problem, which also requires the input of a target topic in addition to a piece of descriptive text. The key reason for proposing the new problem is that in practical applications, we found that useful questions need to be targeted toward some relevant topics. One almost never asks a random question in a conversation. Due to the fact that given a descriptive text, it is often possible to ask many types of questions, generating a question without knowing what it is about is of limited use. To solve the problem, we propose a novel neural network that is able to generate topic-specific questions. One major advantage of this model is that it can be trained directly using a question-answering corpus without requiring any additional annotations like annotating topics in the questions or answers. Experimental results show that our model outperforms the state-of-the-art baseline.", "target": ["Proponiamo una rete neurale che è in grado di generare domande specifiche per argomento.", "Presenta un approccio basato su reti neurali per generare domande specifiche per argomento con la motivazione che le domande relative ad un argomento sono più significative nelle applicazioni pratiche.", "Propone un metodo di generazione basato su argomenti utilizzando una LSTM per estrarre gli argomenti utilizzando una tecnica di codifica a due stadi"]} +{"source": "Brain-Machine Interfaces (BMIs) have recently emerged as a clinically viable option to restore voluntary movements after paralysis. These devices are based on the ability to extract information about movement intent from neural signals recorded using multi-electrode arrays chronically implanted in the motor cortices of the brain. However, the inherent loss and turnover of recorded neurons requires repeated recalibrations of the interface, which can potentially alter the day-to-day user experience. The resulting need for continued user adaptation interferes with the natural, subconscious use of the BMI. Here, we introduce a new computational approach that decodes movement intent from a low-dimensional latent representation of the neural data. We implement various domain adaptation methods to stabilize the interface over significantly long times. This includes Canonical Correlation Analysis used to align the latent variables across days; this method requires prior point-to-point correspondence of the time series across domains. Alternatively, we match the empirical probability distributions of the latent variables across days through the minimization of their Kullback-Leibler divergence. These two methods provide a significant and comparable improvement in the performance of the interface. However, implementation of an Adversarial Domain Adaptation Network trained to match the empirical probability distribution of the residuals of the reconstructed neural signals outperforms the two methods based on latent variables, while requiring remarkably few data points to solve the domain adaptation problem.", "target": ["Implementiamo una rete di adversarial domain adaptation per stabilizzare un'interfaccia Brain-Machine fissa contro i cambiamenti graduali nei segnali neurali registrati.", "Descrive un nuovo approccio per l'interfaccia brain machine impiantata per affrontare il problema della calibrazione e il covariate shift.", "Gli autori definiscono una BMI che utilizza un autoencoder e poi affrontano il problema del data drift nella BMI."]} +{"source": "Neural network-based systems can now learn to locate the referents of words and phrases in images, answer questions about visual scenes, and even execute symbolic instructions as first-person actors in partially-observable worlds. To achieve this so-called grounded language learning, models must overcome certain well-studied learning challenges that are also fundamental to infants learning their first words. While it is notable that models with no meaningful prior knowledge overcome these learning obstacles, AI researchers and practitioners currently lack a clear understanding of exactly how they do so. Here we address this question as a way of achieving a clearer general understanding of grounded language learning, both to inform future research and to improve confidence in model predictions. For maximum control and generality, we focus on a simple neural network-based language learning agent trained via policy-gradient methods to interpret synthetic linguistic instructions in a simulated 3D world. We apply experimental paradigms from developmental psychology to this agent, exploring the conditions under which established human biases and learning effects emerge. We further propose a novel way to visualise and analyse semantic representation in grounded language learning agents that yields a plausible computational account of the observed effects.", "target": ["Analizzare e capire come gli agenti delle reti neurali imparano a capire un linguaggio semplice e grounded", "Gli autori collegano i metodi sperimentali psicologici alla comprensione di come la scatola nera dei metodi di deep learning risolva i problemi.", "Questo articolo presenta un'analisi degli agenti che imparano il linguaggio grounded attraverso reinforcement learning in un ambiente semplice che combina istruzioni verbali con informazioni visive"]} +{"source": "Humans easily recognize object parts and their hierarchical structure by watching how they move; they can then predict how each part moves in the future. In this paper, we propose a novel formulation that simultaneously learns a hierarchical, disentangled object representation and a dynamics model for object parts from unlabeled videos. Our Parts, Structure, and Dynamics (PSD) model learns to, first, recognize the object parts via a layered image representation; second, predict hierarchy via a structural descriptor that composes low-level concepts into a hierarchical structure; and third, model the system dynamics by predicting the future. Experiments on multiple real and synthetic datasets demonstrate that our PSD model works well on all three tasks: segmenting object parts, building their hierarchical structure, and capturing their motion distributions.", "target": ["Imparare le parti degli oggetti, la struttura gerarchica e le dinamiche guardando come si muovono", "Propone un modello di apprendimento unsupervised che impara a separare gli oggetti in parti, prevedere la struttura gerarchica delle parti e, sulla base delle parti separate e della gerarchia, prevedere il movimento."]} +{"source": "A successful application of convolutional architectures is to increase the resolution of single low-resolution images -- a image restoration task called super-resolution (SR). Naturally, SR is of value to resource constrained devices like mobile phones, electronic photograph frames and televisions to enhance image quality. However, SR demands perhaps the most extreme amounts of memory and compute operations of any mainstream vision task known today, preventing SR from being deployed to devices that require them. In this paper, we perform a early systematic study of system resource efficiency for SR, within the context of a variety of architectural and low-precision approaches originally developed for discriminative neural networks. We present a rich set of insights, representative SR architectures, and efficiency trade-offs; for example, the prioritization of ways to compress models to reach a specific memory and computation target and techniques to compact SR models so that they are suitable for DSPs and FPGAs. As a result of doing so, we manage to achieve better and comparable performance with previous models in the existing literature, highlighting the practicality of using existing efficiency techniques in SR tasks. Collectively, we believe these results provides the foundation for further research into the little explored area of resource efficiency for SR.", "target": ["Cerchiamo di comprendere le tecniche efficienti in termini di risorse per la Super-Risoluzione", "L'articolo propone una valutazione empirica dettagliata dei compromessi raggiunti da varie reti neurali convoluzionali sul problema della super risoluzione.", "Questo articolo ha proposto di migliorare l'efficienza delle risorse del sistema per le reti di super risoluzione."]} +{"source": "Neural networks are known to be a class of highly expressive functions able to fit even random input-output mappings with 100% accuracy. In this work we present properties of neural networks that complement this aspect of expressivity. By using tools from Fourier analysis, we show that deep ReLU networks are biased towards low frequency functions, meaning that they cannot have local fluctuations without affecting their global behavior. Intuitively, this property is in line with the observation that over-parameterized networks find simple patterns that generalize across data samples. We also investigate how the shape of the data manifold affects expressivity by showing evidence that learning high frequencies gets easier with increasing manifold complexity, and present a theoretical understanding of this behavior. Finally, we study the robustness of the frequency components with respect to parameter perturbation, to develop the intuition that the parameters must be finely tuned to express high frequency functions.", "target": ["Studiamo le reti ReLU nel dominio di Fourier e dimostriamo un comportamento particolare.", "Analisi di Fourier delle reti ReLU, trovando che sono biased verso l'apprendimento a bassa frequenza", "Questo articolo ha contributi teorici ed empirici sul tema dei coefficienti di Fourier delle reti neurali"]} +{"source": "Instance embeddings are an efficient and versatile image representation that facilitates applications like recognition, verification, retrieval, and clustering. Many metric learning methods represent the input as a single point in the embedding space. Often the distance between points is used as a proxy for match confidence. However, this can fail to represent uncertainty which can arise when the input is ambiguous, e.g., due to occlusion or blurriness. This work addresses this issue and explicitly models the uncertainty by “hedging” the location of each input in the embedding space. We introduce the hedged instance embedding (HIB) in which embeddings are modeled as random variables and the model is trained under the variational information bottleneck principle (Alemi et al., 2016; Achille & Soatto, 2018). Empirical results on our new N-digit MNIST dataset show that our method leads to the desired behavior of “hedging its bets” across the embedding space upon encountering ambiguous inputs. This results in improved performance for image matching and classification tasks, more structure in the learned embedding space, and an ability to compute a per-exemplar uncertainty measure which is correlated with downstream performance.", "target": ["L'articolo propone di usare distribuzioni di probabilità invece dei punti per task di embedding di istanze come il riconoscimento e la verifica.", "Il paper propone un'alternativa all'attuale embedding di punti e una tecnica per addestrarli.", "L'articolo propone un modello che utilizza gli embedding di incertezze per estendere il deep learning alle applicazioni bayesiane"]} +{"source": "Convolution neural networks typically consist of many convolutional layers followed by several fully-connected layers. While convolutional layers map between high-order activation tensors, the fully-connected layers operate on flattened activation vectors. Despite its success, this approach has notable drawbacks. Flattening discards the multi-dimensional structure of the activations, and the fully-connected layers require a large number of parameters. We present two new techniques to address these problems. First, we introduce tensor contraction layers which can replace the ordinary fully-connected layers in a neural network. Second, we introduce tensor regression layers, which express the output of a neural network as a low-rank multi-linear mapping from a high-order activation tensor to the softmax layer. Both the contraction and regression weights are learned end-to-end by backpropagation. By imposing low rank on both, we use significantly fewer parameters. Experiments on the ImageNet dataset show that applied to the popular VGG and ResNet architectures, our methods significantly reduce the number of parameters in the fully connected layers (about 65% space savings) while negligibly impacting accuracy.", "target": ["Proponiamo i layer di contrazione e di regressione dei tensori a basso rango per preservare e sfruttare la struttura multi-lineare in tutta la rete, ottenendo un enorme risparmio di spazio con poco o nessun impatto sulle prestazioni.", "Questo articolo propone nuove architetture di layer di reti neurali utilizzando una rappresentazione a basso rango di tensori", "Questo articolo incorpora la decomposizione e la regressione tensoriale nelle CNN utilizzando un nuovo layer di regressione tensoriale."]} +{"source": "We explore ways of incorporating bilingual dictionaries to enable semi-supervised neural machine translation. Conventional back-translation methods have shown success in leveraging target side monolingual data. However, since the quality of back-translation models is tied to the size of the available parallel corpora, this could adversely impact the synthetically generated sentences in a low resource setting. We propose a simple data augmentation technique to address both this shortcoming. We incorporate widely available bilingual dictionaries that yield word-by-word translations to generate synthetic sentences. This automatically expands the vocabulary of the model while maintaining high quality content. Our method shows an appreciable improvement in performance over strong baselines.", "target": ["Usiamo i dizionari bilingui per data augmentation per la traduzione automatica neurale", "Questo articolo studia l'uso di dizionari bilingui per creare fonti sintetiche per i dati monolinguali lato target per migliorare i modelli di NMT addestrati con piccole quantità di dati paralleli."]} +{"source": "Rewards are sparse in the real world and most of today's reinforcement learning algorithms struggle with such sparsity. One solution to this problem is to allow the agent to create rewards for itself - thus making rewards dense and more suitable for learning. In particular, inspired by curious behaviour in animals, observing something novel could be rewarded with a bonus. Such bonus is summed up with the real task reward - making it possible for RL algorithms to learn from the combined reward. We propose a new curiosity method which uses episodic memory to form the novelty bonus. To determine the bonus, the current observation is compared with the observations in memory. Crucially, the comparison is done based on how many environment steps it takes to reach the current observation from those in memory - which incorporates rich information about environment dynamics. This allows us to overcome the known \"couch-potato\" issues of prior work - when the agent finds a way to instantly gratify itself by exploiting actions which lead to hardly predictable consequences. We test our approach in visually rich 3D environments in ViZDoom, DMLab and MuJoCo. In navigational tasks from ViZDoom and DMLab, our agent outperforms the state-of-the-art curiosity method ICM. In MuJoCo, an ant equipped with our curiosity module learns locomotion out of the first-person-view curiosity only. The code is available at https://github.com/google-research/episodic-curiosity/.", "target": ["Proponiamo un nuovo modello di curiosity basato sulla memoria episodica e le idee di raggiungibilità che ci permette di superare i noti problemi di \"couch-potato\" dei lavori precedenti.", "Propone di dare bonus di esplorazione negli algoritmi RL dando bonus più grandi alle osservazioni che sono più lontane in termini di passi nell'ambiente.", "Gli autori propongono un bonus di esplorazione che ha lo scopo di aiutare nei problemi RL a reward sparsa e considerano molti esperimenti su ambienti 3D complessi"]} +{"source": "We introduce a new dataset of logical entailments for the purpose of measuring models' ability to capture and exploit the structure of logical expressions against an entailment prediction task. We use this task to compare a series of architectures which are ubiquitous in the sequence-processing literature, in addition to a new model class---PossibleWorldNets---which computes entailment as a ``convolution over possible worlds''. Results show that convolutional networks present the wrong inductive bias for this class of problems relative to LSTM RNNs, tree-structured neural networks outperform LSTM RNNs due to their enhanced ability to exploit the syntax of logic, and PossibleWorldNets outperform all benchmarks.", "target": ["Introduciamo un nuovo dataset di implicazioni logiche allo scopo di misurare la capacità dei modelli di catturare e sfruttare la struttura delle espressioni logiche in un task di entailment prediction.", "L'articolo propone un nuovo modello per utilizzare deep model per rilevare l'implicazione logica come prodotto di funzioni continue su mondi possibili.", "Propone un nuovo modello progettato per machine learning con logical entailment prediction."]} +{"source": "Deep convolutional neural network (DCNN) based supervised learning is a widely practiced approach for large-scale image classification. However, retraining these large networks to accommodate new, previously unseen data demands high computational time and energy requirements. Also, previously seen training samples may not be available at the time of retraining. We propose an efficient training methodology and incrementally growing a DCNN to allow new classes to be learned while sharing part of the base network. Our proposed methodology is inspired by transfer learning techniques, although it does not forget previously learned classes. An updated network for learning new set of classes is formed using previously learned convolutional layers (shared from initial part of base network) with addition of few newly added convolutional kernels included in the later layers of the network. We evaluated the proposed scheme on several recognition applications. The classification accuracy achieved by our approach is comparable to the regular incremental learning approach (where networks are updated with new training samples only, without any network sharing).", "target": ["L'articolo riguarda una nuova metodologia ad alta efficienza energetica per incremental learning", "Propone una procedura per l'incremental learning come transfer learning.", "L'articolo presenta un metodo per addestrare deep convolutional neural network in modo incrementale, in cui i dati sono disponibili in piccole batch per un periodo di tempo.", "Presenta un approccio al class-incremental learning utilizzando deep network, proponendo tre diverse strategie di apprendimento nell'approccio finale/migliore."]} +{"source": "Recurrent neural networks (RNNs) are widely used to model sequential data but their non-linear dependencies between sequence elements prevent parallelizing training over sequence length. We show the training of RNNs with only linear sequential dependencies can be parallelized over the sequence length using the parallel scan algorithm, leading to rapid training on long sequences even with small minibatch size. We develop a parallel linear recurrence CUDA kernel and show that it can be applied to immediately speed up training and inference of several state of the art RNN architectures by up to 9x. We abstract recent work on linear RNNs into a new framework of linear surrogate RNNs and develop a linear surrogate model for the long short-term memory unit, the GILR-LSTM, that utilizes parallel linear recurrence. We extend sequence learning to new extremely long sequence regimes that were previously out of reach by successfully training a GILR-LSTM on a synthetic sequence classification task with a one million timestep dependency.", "target": ["Utilizzare la scansione parallela per parallelizzare le reti neurali ricorrenti lineari. Si addestra il modello sulla lunghezza di 1 milione di dipendenze", "Propone di accelerare le RNN applicando il metodo di Blelloch.", "Gli autori propongono un algoritmo parallelo per le Linear Surrogate RNN, che produce uno speedup rispetto alle implementazioni esistenti di Quasi-RNN, SRU e LSTM."]} +{"source": "Neural text generation models are often autoregressive language models or seq2seq models. Neural autoregressive and seq2seq models that generate text by sampling words sequentially, with each word conditioned on the previous model, are state-of-the-art for several machine translation and summarization benchmarks. These benchmarks are often defined by validation perplexity even though this is not a direct measure of sample quality. Language models are typically trained via maximum likelihood and most often with teacher forcing. Teacher forcing is well-suited to optimizing perplexity but can result in poor sample quality because generating text requires conditioning on sequences of words that were never observed at training time. We propose to improve sample quality using Generative Adversarial Network (GANs), which explicitly train the generator to produce high quality samples and have shown a lot of success in image generation. GANs were originally to designed to output differentiable values, so discrete language generation is challenging for them. We introduce an actor-critic conditional GAN that fills in missing text conditioned on the surrounding context. We show qualitatively and quantitatively, evidence that this produces more realistic text samples compared to a maximum likelihood trained model.", "target": ["GAN per il linguaggio naturale per il riempimento dello spazio vuoto", "Questo articolo propone di generare testo usando le GAN.", "Generazione di sample di testo utilizzando le GAN e un meccanismo per riempire le parole mancanti condizionato dal testo circostante"]} +{"source": "Parametric texture models have been applied successfully to synthesize artificial images. Psychophysical studies show that under defined conditions observers are unable to differentiate between model-generated and original natural textures. In industrial applications the reverse case is of interest: a texture analysis system should decide if human observers are able to discriminate between a reference and a novel texture. For example, in case of inspecting decorative surfaces the de- tection of visible texture anomalies without any prior knowledge is required. Here, we implemented a human-vision-inspired novelty detection approach. Assuming that the features used for texture synthesis are important for human texture percep- tion, we compare psychophysical as well as learnt texture representations based on activations of a pretrained CNN in a novelty detection scenario. Additionally, we introduce a novel objective function to train one-class neural networks for novelty detection and compare the results to standard one-class SVM approaches. Our experiments clearly show the differences between human-vision-inspired texture representations and learnt features in detecting visual anomalies. Based on a dig- ital print inspection scenario we show that psychophysical texture representations are able to outperform CNN-encoded features.", "target": ["Confronto tra le rappresentazioni delle texture psicofisiche e quelle codificate dalla CNN in un'applicazione di novelty detection con rete neurale ad una classe.", "Questo articolo si concentra su novelty detection e mostra che le rappresentazioni psicofisiche possono superare le feature del VGG-encoder in una parte di questo task", "Questo articolo considera il rilevamento delle anomalie nelle texture e propone una loss originale.", "Propone il training di due rilevatori di anomalie da tre modelli diversi per rilevare anomalie percettive nelle texture visive."]} +{"source": "In representation learning (RL), how to make the learned representations easy to interpret and less overfitted to training data are two important but challenging issues. To address these problems, we study a new type of regularization approach that encourages the supports of weight vectors in RL models to have small overlap, by simultaneously promoting near-orthogonality among vectors and sparsity of each vector. We apply the proposed regularizer to two models: neural networks (NNs) and sparse coding (SC), and develop an efficient ADMM-based algorithm for regularized SC. Experiments on various datasets demonstrate that weight vectors learned under our regularizer are more interpretable and have better generalization performance.", "target": ["Proponiamo un nuovo tipo di approccio di regolarizzazione che incoraggia la non sovrapposizione nel representation learning, al fine di migliorare l'interpretabilità e ridurre l'overfitting.", "L'articolo introduce un regolarizzatore di matrice per indurre simultaneamente sia la sparsità che l'ortogonalità approssimativa.", "L'articolo studia un metodo di regolarizzazione per promuovere la sparsità e ridurre la sovrapposizione tra i supporti dei vettori dei pesi nelle rappresentazioni apprese per migliorare l'interpretabilità ed evitare l'overfitting", "L'articolo ha proposto un nuovo approccio di regolarizzazione che incoraggia simultaneamente i vettori dei pesi (W) ad essere sparsi e ortogonali tra loro."]} +{"source": "Successful recurrent models such as long short-term memories (LSTMs) and gated recurrent units (GRUs) use \\emph{ad hoc} gating mechanisms. Empirically these models have been found to improve the learning of medium to long term temporal dependencies and to help with vanishing gradient issues. \t We prove that learnable gates in a recurrent model formally provide \\emph{quasi-invariance to general time transformations} in the input data. We recover part of the LSTM architecture from a simple axiomatic approach. \t This result leads to a new way of initializing gate biases in LSTMs and GRUs. Experimentally, this new \\emph{chrono initialization} is shown to greatly improve learning of long term dependencies, with minimal implementation effort.", "target": ["Dimostra che i meccanismi di gating forniscono invarianza alle trasformazioni temporali. Introduce e testa una nuova inizializzazione per le LSTM da questa intuizione.", "Il paper collega il design delle reti ricorrenti e il suo effetto su come la rete reagisce alle trasformazioni del tempo, e usa questo per sviluppare un semplice schema di inizializzazione dei bias."]} +{"source": "Teaching plays a very important role in our society, by spreading human knowledge and educating our next generations. A good teacher will select appropriate teaching materials, impact suitable methodologies, and set up targeted examinations, according to the learning behaviors of the students. In the field of artificial intelligence, however, one has not fully explored the role of teaching, and pays most attention to machine \\emph{learning}. In this paper, we argue that equal attention, if not more, should be paid to teaching, and furthermore, an optimization framework (instead of heuristics) should be used to obtain good teaching strategies. We call this approach ``learning to teach''. In the approach, two intelligent agents interact with each other: a student model (which corresponds to the learner in traditional machine learning algorithms), and a teacher model (which determines the appropriate data, loss function, and hypothesis space to facilitate the training of the student model). The teacher model leverages the feedback from the student model to optimize its own teaching strategies by means of reinforcement learning, so as to achieve teacher-student co-evolution. To demonstrate the practical value of our proposed approach, we take the training of deep neural networks (DNN) as an example, and show that by using the learning to teach techniques, we are able to use much less training data and fewer iterations to achieve almost the same accuracy for different kinds of DNN models (e.g., multi-layer perceptron, convolutional neural networks and recurrent neural networks) under various machine learning tasks (e.g., image classification and text understanding).", "target": ["Proponiamo e verifichiamo l'efficacia del \"learning to teach\", un nuovo framework per guidare automaticamente il processo di apprendimento automatico.", "Questo articolo si concentra sul \"machine teaching\" e propone di sfruttare reinforcement learning definendo la reward come la velocità di apprendimento dell'allievo e usando il policy gradient per aggiornare i parametri dell'insegnante", "Gli autori definiscono un modello di deep learning composto da quattro componenti: un modello di studente, un modello di insegnante, una loss e un dataset.", "Suggerisce un framework di \"learning to teach\", corrispondente alle scelte sui dati presentati al learner."]} +{"source": "We present DL2, a system for training and querying neural networks with logical constraints. The key idea is to translate these constraints into a differentiable loss with desirable mathematical properties and to then either train with this loss in an iterative manner or to use the loss for querying the network for inputs subject to the constraints. We empirically demonstrate that DL2 is effective in both training and querying scenarios, across a range of constraints and data sets.", "target": ["Una loss differenziabile per i vincoli logici per il training e il querying delle reti neurali.", "Un framework per trasformare il querying sui parametri e coppie di input-ouput alle reti neurali in funzioni di loss differenziabili e un linguaggio dichiarativo associato per specificare queste query", "Questo articolo affronta il problema di combinare approcci logici con reti neurali traducendo una formula logica in una loss non negativa per una rete neurale."]} +{"source": "Genetic algorithms have been widely used in many practical optimization problems. Inspired by natural selection, operators, including mutation, crossover and selection, provide effective heuristics for search and black-box optimization. However, they have not been shown useful for deep reinforcement learning, possibly due to the catastrophic consequence of parameter crossovers of neural networks. Here, we present Genetic Policy Optimization (GPO), a new genetic algorithm for sample-efficient deep policy optimization. GPO uses imitation learning for policy crossover in the state space and applies policy gradient methods for mutation. Our experiments on MuJoCo tasks show that GPO as a genetic algorithm is able to provide superior performance over the state-of-the-art policy gradient methods and achieves comparable or higher sample efficiency.", "target": ["Approccio basato su algoritmi genetici per l'ottimizzazione delle policy delle deep neural network", "Gli autori presentano un algoritmo per il training di ensemble di policy network che mescola regolarmente diverse policy nell'ensemble.", "Questo articolo propone un metodo di ottimizzazione delle policy ispirato agli algoritmi genetici, che imita gli operatori di mutazione e crossover sulle policy network."]} +{"source": "To make deep neural networks feasible in resource-constrained environments (such as mobile devices), it is beneficial to quantize models by using low-precision weights. One common technique for quantizing neural networks is the straight-through gradient method, which enables back-propagation through the quantization mapping. Despite its empirical success, little is understood about why the straight-through gradient method works. Building upon a novel observation that the straight-through gradient method is in fact identical to the well-known Nesterov’s dual-averaging algorithm on a quantization constrained optimization problem, we propose a more principled alternative approach, called ProxQuant , that formulates quantized network training as a regularized learning problem instead and optimizes it via the prox-gradient method. ProxQuant does back-propagation on the underlying full-precision vector and applies an efficient prox-operator in between stochastic gradient steps to encourage quantizedness. For quantizing ResNets and LSTMs, ProxQuant outperforms state-of-the-art results on binary quantization and is on par with state-of-the-art on multi-bit quantization. We further perform theoretical analyses showing that ProxQuant converges to stationary points under mild smoothness assumptions, whereas variants such as lazy prox-gradient method can fail to converge in the same setting.", "target": ["Un framework per la quantizzazione del modello usando il metodo del gradiente prossimale, con valutazione empirica e analisi teorica della convergenza.", "Propone il metodo ProxQuant per addestrare reti neurali con pesi quantizzati.", "Propone di risolvere le reti binarie e le sue varianti usando la proximal gradient descent."]} +{"source": "Background: Statistical mechanics results (Dauphin et al. (2014); Choromanska et al. (2015)) suggest that local minima with high error are exponentially rare in high dimensions. However, to prove low error guarantees for Multilayer Neural Networks (MNNs), previous works so far required either a heavily modified MNN model or training method, strong assumptions on the labels (e.g., “near” linear separability), or an unrealistically wide hidden layer with \\Omega\\(N) units. Results: We examine a MNN with one hidden layer of piecewise linear units, a single output, and a quadratic loss. We prove that, with high probability in the limit of N\\rightarrow\\infty datapoints, the volume of differentiable regions of the empiric loss containing sub-optimal differentiable local minima is exponentially vanishing in comparison with the same volume of global minima, given standard normal input of dimension d_0=\\tilde{\\Omega}(\\sqrt{N}), and a more realistic number of d_1=\\tilde{\\Omega}(N/d_0) hidden units. We demonstrate our results numerically: for example, 0% binary classification training error on CIFAR with only N/d_0 = 16 hidden neurons.", "target": ["I minimi locali \"cattivi\" svaniscono in una rete neurale multilayer: una prova con ipotesi più ragionevoli di prima", "Nelle reti con un solo hidden layer, il volume dei minimi locali subottimali diminuisce esponenzialmente rispetto ai minimi globali.", "Questo articolo mira a rispondere al perché gli algoritmi standard basati su SGD su rete neurale convergono a soluzioni \"buone\"."]} +{"source": "Deep neural networks are vulnerable to adversarial examples, which can mislead classifiers by adding imperceptible perturbations. An intriguing property of adversarial examples is their good transferability, making black-box attacks feasible in real-world applications. Due to the threat of adversarial attacks, many methods have been proposed to improve the robustness, and several state-of-the-art defenses are shown to be robust against transferable adversarial examples. In this paper, we identify the attention shift phenomenon, which may hinder the transferability of adversarial examples to the defense models. It indicates that the defenses rely on different discriminative regions to make predictions compared with normally trained models. Therefore, we propose an attention-invariant attack method to generate more transferable adversarial examples. Extensive experiments on the ImageNet dataset validate the effectiveness of the proposed method. Our best attack fools eight state-of-the-art defenses at an 82% success rate on average based only on the transferability, demonstrating the insecurity of the defense techniques.", "target": ["Proponiamo un metodo di attacco invariante all'attention per generare adversarial example più trasferibili per gli attacchi black-box, che possono ingannare le difese allo stato dell'arte con un alto tasso di successo.", "L'articolo propone un nuovo modo di superare lo stato dell'arte delle difese contro gli adversarial attack alle CNN.", "Questo articolo suggerisce che l'attention shift è una proprietà chiave dietro il fallimento degli adversarial attack da trasferire e propone un metodo di attacco invariante all'attention"]} +{"source": "We present Merged-Averaged Classifiers via Hashing (MACH) for $K$-classification with large $K$. Compared to traditional one-vs-all classifiers that require $O(Kd)$ memory and inference cost, MACH only need $O(d\\log{K})$ memory while only requiring $O(K\\log{K} + d\\log{K})$ operation for inference. MACH is the first generic $K$-classification algorithm, with provably theoretical guarantees, which requires $O(\\log{K})$ memory without any assumption on the relationship between classes. MACH uses universal hashing to reduce classification with a large number of classes to few independent classification task with very small (constant) number of classes. We provide theoretical quantification of accuracy-memory tradeoff by showing the first connection between extreme classification and heavy hitters. With MACH we can train ODP dataset with 100,000 classes and 400,000 features on a single Titan X GPU (12GB), with the classification accuracy of 19.28\\%, which is the best-reported accuracy on this dataset. Before this work, the best performing baseline is a one-vs-all classifier that requires 40 billion parameters (320 GB model size) and achieves 9\\% accuracy. In contrast, MACH can achieve 9\\% accuracy with 480x reduction in the model size (of mere 0.6GB). With MACH, we also demonstrate complete training of fine-grained imagenet dataset (compressed size 104GB), with 21,000 classes, on a single GPU.", "target": ["Come allenare 100.000 classi su una singola GPU", "Propone un efficiente metodo di hashing MACH per l'approssimazione di softmax nel contesto di un grande spazio di output, che risparmia sia memoria che computazione.", "Un metodo per lo schema di classificazione per problemi che coinvolgono un gran numero di classi in un ambiente multiclasse dimostrato sui dataset ODP e Imagenet-21K", "L'articolo presenta uno schema basato sull'hashing per ridurre la memoria e il tempo di calcolo per la classificazione K-way quando K è grande"]} +{"source": "Gradient-based optimization is the foundation of deep learning and reinforcement learning. Even when the mechanism being optimized is unknown or not differentiable, optimization using high-variance or biased gradient estimates is still often the best strategy. We introduce a general framework for learning low-variance, unbiased gradient estimators for black-box functions of random variables, based on gradients of a learned function. These estimators can be jointly trained with model parameters or policies, and are applicable in both discrete and continuous settings. We give unbiased, adaptive analogs of state-of-the-art reinforcement learning methods such as advantage actor-critic. We also demonstrate this framework for training discrete latent-variable models.", "target": ["Presentiamo un metodo generale per la stima unbiased dei gradienti di funzioni black-box di variabili casuali. Applichiamo questo metodo all'inferenza variazionale discreta e al reinforcement learning.", "Suggerisce un nuovo approccio per eseguire la gradient descent per l'ottimizzazione blackbox o il training di modelli discreti a variabili latenti."]} +{"source": "Do GANS (Generative Adversarial Nets) actually learn the target distribution? The foundational paper of Goodfellow et al. (2014) suggested they do, if they were given sufficiently large deep nets, sample size, and computation time. A recent theoretical analysis in Arora et al. (2017) raised doubts whether the same holds when discriminator has bounded size. It showed that the training objective can approach its optimum value even if the generated distribution has very low support. In other words, the training objective is unable to prevent mode collapse. The current paper makes two contributions. (1) It proposes a novel test for estimating support size using the birthday paradox of discrete probability. Using this evidence is presented that well-known GANs approaches do learn distributions of fairly low support. (2) It theoretically studies encoder-decoder GANs architectures (e.g., BiGAN/ALI), which were proposed to learn more meaningful features via GANs, and consequently to also solve the mode-collapse issue. Our result shows that such encoder-decoder training objectives also cannot guarantee learning of the full distribution because they cannot prevent serious mode collapse. More seriously, they cannot prevent learning meaningless codes for data, contrary to usual intuition.", "target": ["Proponiamo uno stimatore della dimensione del supporto della distribuzione appresa dalle GAN per dimostrare che soffrono davvero di mode collapse, e dimostriamo che anche le GAN encoder-decoder non evitano il problema.", "L'articolo tenta di stimare sperimentalmente la dimensione del supporto delle soluzioni prodotte dalle GAN standard.", "Questo articolo propone un nuovo test intelligente basato sul birthday paradox per misurare la diversità nel sample generato, con i risultati sperimentali che indicano che il mode collapse è forte in un certo numero di modelli generativi allo stato dell'arte.", "L'articolo usa il birthday paradox per mostrare che alcune architetture GAN generano distribuzioni con un supporto abbastanza basso."]} +{"source": "Due to their complex nature, it is hard to characterize the ways in which machine learning models can misbehave or be exploited when deployed. Recent work on adversarial examples, i.e. inputs with minor perturbations that result in substantially different model predictions, is helpful in evaluating the robustness of these models by exposing the adversarial scenarios where they fail. However, these malicious perturbations are often unnatural, not semantically meaningful, and not applicable to complicated domains such as language. In this paper, we propose a framework to generate natural and legible adversarial examples that lie on the data manifold, by searching in semantic space of dense and continuous data representation, utilizing the recent advances in generative adversarial networks. We present generated adversaries to demonstrate the potential of the proposed approach for black-box classifiers for a wide range of applications such as image classification, textual entailment, and machine translation. We include experiments to show that the generated adversaries are natural, legible to humans, and useful in evaluating and analyzing black-box classifiers.", "target": ["Proponiamo un framework per generare avversari naturali contro i classificatori black-box sia per domini visivi che testuali, facendo la ricerca di avversari nello spazio semantico latente.", "Suggerisce un metodo per la creazione di adversarial example semantici.", "Propone un framework per generare adversarial example naturali cercando avversari in uno spazio latente di rappresentazione di dati densi e continui"]} +{"source": "Kronecker-factor Approximate Curvature (Martens & Grosse, 2015) (K-FAC) is a 2nd-order optimization method which has been shown to give state-of-the-art performance on large-scale neural network optimization tasks (Ba et al., 2017). It is based on an approximation to the Fisher information matrix (FIM) that makes assumptions about the particular structure of the network and the way it is parameterized. The original K-FAC method was applicable only to fully-connected networks, although it has been recently extended by Grosse & Martens (2016) to handle convolutional networks as well. In this work we extend the method to handle RNNs by introducing a novel approximation to the FIM for RNNs. This approximation works by modelling the covariance structure between the gradient contributions at different time-steps using a chain-structured linear Gaussian graphical model, summing the various cross-covariances, and computing the inverse in closed form. We demonstrate in experiments that our method significantly outperforms general purpose state-of-the-art optimizers like SGD with momentum and Adam on several challenging RNN training tasks.", "target": ["Estendiamo il metodo K-FAC alle RNN sviluppando una nuova famiglia di approssimazioni di Fisher.", "Gli autori estendono il metodo K-FAC alle RNN e presentano 3 modi di approssimare F, mostrando risultati di ottimizzazione su 3 dataset, che superano ADAM sia nel numero di aggiornamenti che nel tempo di calcolo.", "Propone di estendere il metodo di ottimizzazione di Kronecker-factor Appropriate Curvature al setting delle reti neurali ricorrenti.", "Gli autori presentano un metodo del secondo ordine che è specificamente progettato per le RNN"]} +{"source": "Bayesian inference is known to provide a general framework for incorporating prior knowledge or specific properties into machine learning models via carefully choosing a prior distribution. In this work, we propose a new type of prior distributions for convolutional neural networks, deep weight prior (DWP), that exploit generative models to encourage a specific structure of trained convolutional filters e.g., spatial correlations of weights. We define DWP in the form of an implicit distribution and propose a method for variational inference with such type of implicit priors. In experiments, we show that DWP improves the performance of Bayesian neural networks when training data are limited, and initialization of weights with samples from DWP accelerates training of conventional convolutional neural networks.", "target": ["Il modello generativo per i kernel delle reti neurali convoluzionali, che agisce come una distribuzione prior durante il training su nuovi dataset.", "Un metodo per modellare le reti neurali convoluzionali usando un metodo di Bayes.", "Propone il \"deep weight prior\": l'idea è quella di ottenere un prior su un dataset ausiliario e poi usare quel prior sui filtri CNN per avviare l'inferenza per un dataset di interesse.", "Questo articolo esplora l'apprendimento di prior informativi per modelli di reti neurali convoluzionali con domini simili utilizzando autoencoder per ottenere un prior espressivo sui pesi filtrati delle reti addestrate."]} +{"source": "The high dimensionality of hyperspectral imaging forces unique challenges in scope, size and processing requirements. Motivated by the potential for an in-the-field cell sorting detector, we examine a Synechocystis sp. PCC 6803 dataset wherein cells are grown alternatively in nitrogen rich or deplete cultures. We use deep learning techniques to both successfully classify cells and generate a mask segmenting the cells/condition from the background. Further, we use the classification accuracy to guide a data-driven, iterative feature selection method, allowing the design neural networks requiring 90% fewer input features with little accuracy degradation.", "target": ["Abbiamo applicato tecniche di deep learning alla segmentazione iperspettrale delle immagini e al sampling iterativo delle feature.", "Propone uno schema greedy per selezionare un sottoinsieme di feature spettrali altamente correlate in un task di classificazione.", "L'articolo esplora l'uso delle reti neurali per la classificazione e la segmentazione dell'imaging iperspettrale delle cellule.", "Classificazione delle cellule e implementazione della segmentazione cellulare basata su tecniche di deep learning con riduzione delle feature di input"]} +{"source": "Data Interpretation is an important part of Quantitative Aptitude exams and requires an individual to answer questions grounded in plots such as bar charts, line graphs, scatter plots, \\textit{etc}. Recently, there has been an increasing interest in building models which can perform this task by learning from datasets containing triplets of the form \\{plot, question, answer\\}. Two such datasets have been proposed in the recent past which contain plots generated from synthetic data with limited (i) $x-y$ axes variables (ii) question templates and (iii) answer vocabulary and hence do not adequately capture the challenges posed by this task. To overcome these limitations of existing datasets, we introduce a new dataset containing $9.7$ million question-answer pairs grounded over $270,000$ plots with three main differentiators. First, the plots in our dataset contain a wide variety of realistic $x$-$y$ variables such as CO2 emission, fertility rate, \\textit{etc. } extracted from real word data sources such as World Bank, government sites, \\textit{etc}. Second, the questions in our dataset are more complex as they are based on templates extracted from interesting questions asked by a crowd of workers using a fraction of these plots. Lastly, the answers in our dataset are not restricted to a small vocabulary and a large fraction of the answers seen at test time are not present in the training vocabulary. As a result, existing models for Visual Question Answering which largely use end-to-end models in a multi-class classification framework cannot be used for this task. We establish initial results on this dataset and emphasize the complexity of the task using a multi-staged modular pipeline with various sub-components to (i) extract relevant data from the plot and convert it to a semi-structured table (ii) combine the question with this table and use compositional semantic parsing to arrive at a logical form from which the answer can be derived. We believe that such a modular framework is the best way to go forward as it would enable the research community to independently make progress on all the sub-tasks involved in plot question answering.", "target": ["Abbiamo creato un nuovo dataset per l'interpretazione dei dati su grafico e proponiamo anche una baseline per lo stesso.", "Gli autori propongono una pipeline per risolvere il problema DIP che comporta l'apprendimento da dataset contenenti terzine della forma {grafico, domanda, risposta}", "Propone un algoritmo che può interpretare i dati mostrati nei grafici scientifici."]} +{"source": "Learning to predict complex time-series data is a fundamental challenge in a range of disciplines including Machine Learning, Robotics, and Natural Language Processing. Predictive State Recurrent Neural Networks (PSRNNs) (Downey et al.) are a state-of-the-art approach for modeling time-series data which combine the benefits of probabilistic filters and Recurrent Neural Networks into a single model. PSRNNs leverage the concept of Hilbert Space Embeddings of distributions (Smola et al.) to embed predictive states into a Reproducing Kernel Hilbert Space, then estimate, predict, and update these embedded states using Kernel Bayes Rule. Practical implementations of PSRNNs are made possible by the machinery of Random Features, where input features are mapped into a new space where dot products approximate the kernel well. Unfortunately PSRNNs often require a large number of RFs to obtain good results, resulting in large models which are slow to execute and slow to train. Orthogonal Random Features (ORFs) (Choromanski et al.) is an improvement on RFs which has been shown to decrease the number of RFs required for pointwise kernel approximation. Unfortunately, it is not clear that ORFs can be applied to PSRNNs, as PSRNNs rely on Kernel Ridge Regression as a core component of their learning algorithm, and the theoretical guarantees of ORF do not apply in this setting. In this paper, we extend the theory of ORFs to Kernel Ridge Regression and show that ORFs can be used to obtain Orthogonal PSRNNs (OPSRNNs), which are smaller and faster than PSRNNs. In particular, we show that OPSRNN models clearly outperform LSTMs and furthermore, can achieve accuracy similar to PSRNNs with an order of magnitude smaller number of features needed.", "target": ["Migliorare le reti neurali ricorrenti predictive state tramite feature casuali ortogonali", "Propone di migliorare le prestazioni delle reti neurali ricorrenti predictive state considerando le feature casuali ortogonali.", "L'articolo affronta il problema del training di reti neurali ricorrenti predictive state e apporta due contributi."]} +{"source": "Training deep neural networks requires many training samples, but in practice training labels are expensive to obtain and may be of varying quality, as some may be from trusted expert labelers while others might be from heuristics or other sources of weak supervision such as crowd-sourcing. This creates a fundamental quality- versus-quantity trade-off in the learning process. Do we learn from the small amount of high-quality data or the potentially large amount of weakly-labeled data? We argue that if the learner could somehow know and take the label-quality into account when learning the data representation, we could get the best of both worlds. To this end, we propose “fidelity-weighted learning” (FWL), a semi-supervised student- teacher approach for training deep neural networks using weakly-labeled data. FWL modulates the parameter updates to a student network (trained on the task we care about) on a per-sample basis according to the posterior confidence of its label-quality estimated by a teacher (who has access to the high-quality labels). Both student and teacher are learned from the data. We evaluate FWL on two tasks in information retrieval and natural language processing where we outperform state-of-the-art alternative semi-supervised methods, indicating that our approach makes better use of strong and weak labels, and leads to better task-dependent data representations.", "target": ["Proponiamo il Fidelity-weighted Learning, un approccio semi-supervised teacher-student per il training delle reti neurali usando dati weakly-labeled.", "Questo articolo suggerisce un approccio per l'apprendimento con supervisione debole utilizzando un dataset pulito e uno rumoroso e assumendo un teacher e una rete di student", "L'articolo tenta di addestrare modelli di deep neural network con pochi sample di training annotati.", "Gli autori propongono un approccio per il training di modelli di deep learning per situazioni in cui non ci sono abbastanza dati annotati affidabili."]} +{"source": "Online learning has attracted great attention due to the increasing demand for systems that have the ability of learning and evolving. When the data to be processed is also high dimensional and dimension reduction is necessary for visualization or prediction enhancement, online dimension reduction will play an essential role. The purpose of this paper is to propose new online learning approaches for supervised dimension reduction. Our first algorithm is motivated by adapting the sliced inverse regression (SIR), a pioneer and effective algorithm for supervised dimension reduction, and making it implementable in an incremental manner. The new algorithm, called incremental sliced inverse regression (ISIR), is able to update the subspace of significant factors with intrinsic lower dimensionality fast and efficiently when new observations come in. We also refine the algorithm by using an overlapping technique and develop an incremental overlapping sliced inverse regression (IOSIR) algorithm. We verify the effectiveness and efficiency of both algorithms by simulations and real data applications.", "target": ["Abbiamo proposto due nuovi approcci, la regressione inversa incrementale sliced e la regressione inversa incrementale sliced e overlapping, per implementare la dimension reduction supervisionata in un setting di online learning.", "Studia il problema della dimension reduction sufficiente e propone un algoritmo di regressione inversa incrementale sliced.", "Questo articolo propone un algoritmo di online learning per dimension reduction supervisionata, chiamato regressione inversa incrementale sliced"]} +{"source": "This paper presents a storage-efficient learning model titled Recursive Binary Neural Networks for embedded and mobile devices having a limited amount of on-chip data storage such as hundreds of kilo-Bytes. The main idea of the proposed model is to recursively recycle data storage of weights (parameters) during training. This enables a device with a given storage constraint to train and instantiate a neural network classifier with a larger number of weights on a chip, achieving better classification accuracy. Such efficient use of on-chip storage reduces off-chip storage accesses, improving energy-efficiency and speed of training. We verified the proposed training model with deep and convolutional neural network classifiers on the MNIST and voice activity detection benchmarks. For the deep neural network, our model achieves data storage requirement of as low as 2 bits/weight, whereas the conventional binary neural network learning models require data storage of 8 to 32 bits/weight. With the same amount of data storage, our model can train a bigger network having more weights, achieving 1% less test error than the conventional binary neural network learning model. To achieve the similar classification error, the conventional binary neural network model requires 4× more data storage for weights than our proposed model. For the convolution neural network classifier, the proposed model achieves 2.4% less test error for the same on-chip storage or 6× storage savings to achieve the similar accuracy.", "target": ["Proponiamo un modello di apprendimento che permette alla DNN di imparare con solo 2 bit/peso, che è particolarmente utile per l'apprendimento on-device", "Propone un metodo per discretizzare una NN in modo incrementale per migliorare la memoria e le prestazioni."]} +{"source": "Within-class variation in a high-dimensional dataset can be modeled as being on a low-dimensional manifold due to the constraints of the physical processes producing that variation (e.g., translation, illumination, etc.). We desire a method for learning a representation of the manifolds induced by identity-preserving transformations that can be used to increase robustness, reduce the training burden, and encourage interpretability in machine learning tasks. In particular, what is needed is a representation of the transformation manifold that can robustly capture the shape of the manifold from the input data, generate new points on the manifold, and extend transformations outside of the training domain without significantly increasing the error. Previous work has proposed algorithms to efficiently learn analytic operators (called transport operators) that define the process of transporting one data point on a manifold to another. The main contribution of this paper is to define two transfer learning methods that use this generative manifold representation to learn natural transformations and incorporate them into new data. The first method uses this representation in a novel randomized approach to transfer learning that employs the learned generative model to map out unseen regions of the data space. These results are shown through demonstrations of transfer learning in a data augmentation task for few-shot image classification. The second method use of transport operators for injecting specific transformations into new data examples which allows for realistic image animation and informed data augmentation. These results are shown on stylized constructions using the classic swiss roll data structure and in demonstrations of transfer learning in a data augmentation task for few-shot image classification. We also propose the use of transport operators for injecting transformations into new data examples which allows for realistic image animation.", "target": ["L'apprendimento di operatori di trasporto su manifold forma una rappresentazione preziosa per fare task come il transfer learning.", "Utilizza una framework di dictionary learning per imparare operatori di trasporto su manifold su cifre USPS aumentate.", "L'articolo considera il framework di apprendimento dell'operatore di trasporto su manifold di Culpepper e Olshausen (2009), e lo interpreta come se ottenesse una stima MAP sotto un modello generativo probabilistico."]} +{"source": "The seemingly infinite diversity of the natural world arises from a relatively small set of coherent rules, such as the laws of physics or chemistry. We conjecture that these rules give rise to regularities that can be discovered through primarily unsupervised experiences and represented as abstract concepts. If such representations are compositional and hierarchical, they can be recombined into an exponentially large set of new concepts. This paper describes SCAN (Symbol-Concept Association Network), a new framework for learning such abstractions in the visual domain. SCAN learns concepts through fast symbol association, grounding them in disentangled visual primitives that are discovered in an unsupervised manner. Unlike state of the art multimodal generative model baselines, our approach requires very few pairings between symbols and images and makes no assumptions about the form of symbol representations. Once trained, SCAN is capable of multimodal bi-directional inference, generating a diverse set of image samples from symbolic descriptions and vice versa. It also allows for traversal and manipulation of the implicit hierarchy of visual concepts through symbolic instructions and learnt logical recombination operations. Such manipulations enable SCAN to break away from its training data distribution and imagine novel visual concepts through symbolically instructed recombination of previously learnt concepts.", "target": ["Presentiamo un modello neurale variazionale per l'apprendimento di concetti visivi compositivi guidati dal linguaggio.", "Propone una nuova architettura di rete neurale che impara i concetti degli oggetti combinando un beta-VAE e SCAN.", "Questo articolo introduce un modello basato su VAE per tradurre tra immagini e testo, con la loro rappresentazione latente ben adatta all'applicazione di operazioni simboliche, dando loro un linguaggio più espressivo per il sampling di immagini dal testo.", "Questo articolo propone un nuovo modello chiamato SCAN (Symbol-Concept Association Network) per l'apprendimento gerarchico dei concetti e permette la generalizzazione a nuovi concetti composti da concetti esistenti usando operatori logici."]} +{"source": "Despite much success in many large-scale language tasks, sequence-to-sequence (seq2seq) models have not been an ideal choice for conversational modeling as they tend to generate generic and repetitive responses. In this paper, we propose a Latent Topic Conversational Model (LTCM) that augments the seq2seq model with a neural topic component to better model human-human conversations. The neural topic component encodes information from the source sentence to build a global “topic” distribution over words, which is then consulted by the seq2seq model to improve generation at each time step. The experimental results show that the proposed LTCM can generate more diverse and interesting responses by sampling from its learnt latent representations. In a subjective human evaluation, the judges also confirm that LTCM is the preferred option comparing to competitive baseline models.", "target": ["Latent Topic Conversational Model, un ibrido seq2seq e neural topic model per generare risposte più diverse e interessanti.", "Questo articolo ha proposto la combinazione del topic model e del modello di conversazione seq2seq", "Propone un modello di conversazione con informazioni sull'argomento combinando il modello seq2seq con neural topic model e mostra che il modello proposto supera alcuni modelli baseline seq2seq e altre varianti seq2seq basate su modelli a variabili latenti.", "L'articolo affronta il problema di gestire gli argomenti nei modelli di conversazione e propone un modello che è una combinazione di un neural topic model e un sistema di dialogo basato su seq2seq."]} +{"source": "Most of the existing Graph Neural Networks (GNNs) are the mere extension of the Convolutional Neural Networks (CNNs) to graphs. Generally, they consist of several steps of message passing between the nodes followed by a global indiscriminate feature pooling function. In many data-sets, however, the nodes are unlabeled or their labels provide no information about the similarity between the nodes and the locations of the nodes in the graph. Accordingly, message passing may not propagate helpful information throughout the graph. We show that this conventional approach can fail to learn to perform even simple graph classification tasks. We alleviate this serious shortcoming of the GNNs by making them a two step method. In the first of the proposed approach, a graph embedding algorithm is utilized to obtain a continuous feature vector for each node of the graph. The embedding algorithm represents the graph as a point-cloud in the embedding space. In the second step, the GNN is applied to the point-cloud representation of the graph provided by the embedding method. The GNN learns to perform the given task by inferring the topological structure of the graph encoded in the spatial distribution of the embedded vectors. In addition, we extend the proposed approach to the graph clustering problem and a new architecture for graph clustering is proposed. Moreover, the spatial representation of the graph is utilized to design a graph pooling algorithm. We turn the problem of graph down-sampling into a column sampling problem, i.e., the sampling algorithm selects a subset of the nodes whose feature vectors preserve the spatial distribution of all the feature vectors. We apply the proposed approach to several popular benchmark data-sets and it is shown that the proposed geometrical approach strongly improves the state-of-the-art result for several data-sets. For instance, for the PTC data-set, we improve the state-of-the-art result for more than 22 %.", "target": ["Il problema dell'analisi dei grafi viene trasformato in un problema di analisi delle nuvole di punti.", "Propone una deep GNN network per i problemi di classificazione dei grafi utilizzando il loro layer di pooling graph adaptive.", "Gli autori propongono un metodo per l'apprendimento di rappresentazioni per i grafi"]} +{"source": "Deep neural networks (DNNs) have been found to be vulnerable to adversarial examples resulting from adding small-magnitude perturbations to inputs. Such adversarial examples can mislead DNNs to produce adversary-selected results. Different attack strategies have been proposed to generate adversarial examples, but how to produce them with high perceptual quality and more efficiently requires more research efforts. In this paper, we propose AdvGAN to generate adversarial examples with generative adversarial networks (GANs), which can learn and approximate the distribution of original instances. For AdvGAN, once the generator is trained, it can generate adversarial perturbations efficiently for any instance, so as to potentially accelerate adversarial training as defenses. We apply AdvGAN in both semi-whitebox and black-box attack settings. In semi-whitebox attacks, there is no need to access the original target model after the generator is trained, in contrast to traditional white-box attacks. In black-box attacks, we dynamically train a distilled model for the black-box model and optimize the generator accordingly. Adversarial examples generated by AdvGAN on different target models have high attack success rate under state-of-the-art defenses compared to other attacks. Our attack has placed the first with 92.76% accuracy on a public MNIST black-box attack challenge.", "target": ["Proponiamo di generare un adversarial example basato su adversarial generative network in un setting semi-whitebox e black-box.", "Descrive AdvGAN, una GAN condizionale con loss adversarial, e valuta AdvGAN su setting semi-white box e black box, riportando risultati allo stato dell'arte.", "Questo articolo propone un modo di generare adversarial example che ingannano i sistemi di classificazione e vince la sfida mnist di MadryLab."]} +{"source": "This paper proposes a new model for the rating prediction task in recommender systems which significantly outperforms previous state-of-the art models on a time-split Netflix data set. Our model is based on deep autoencoder with 6 layers and is trained end-to-end without any layer-wise pre-training. We empirically demonstrate that: a) deep autoencoder models generalize much better than the shallow ones, b) non-linear activation functions with negative parts are crucial for training deep models, and c) heavy use of regularization techniques such as dropout is necessary to prevent over-fitting. We also propose a new training algorithm based on iterative output re-feeding to overcome natural sparseness of collaborate filtering. The new algorithm significantly speeds up training and improves model performance. Our code is publicly available.", "target": ["Questo articolo dimostra come addestrare deep autoencoder end-to-end per ottenere risultati SoA sul dataset Netflix time-split.", "Questo articolo presenta un modello di deep autoencoder per la predizione del rating che supera altri approcci allo stato dell'arte sul dataset Netflix prize.", "Propone di utilizzare un deep AE per svolgere task di predizione del rating nei recommender system.", "Gli autori presentano un modello per raccomandazioni Netflix più accurate dimostrando che un deep autoencoder può superare i modelli più complessi basati su RNN che hanno informazioni temporali."]} +{"source": "Recurrent neural networks (RNNs) sequentially process data by updating their state with each new data point, and have long been the de facto choice for sequence modeling tasks. However, their inherently sequential computation makes them slow to train. Feed-forward and convolutional architectures have recently been shown to achieve superior results on some sequence modeling tasks such as machine translation, with the added advantage that they concurrently process all inputs in the sequence, leading to easy parallelization and faster training times. Despite these successes, however, popular feed-forward sequence models like the Transformer fail to generalize in many simple tasks that recurrent models handle with ease, e.g. copying strings or even simple logical inference when the string or formula lengths exceed those observed at training time. We propose the Universal Transformer (UT), a parallel-in-time self-attentive recurrent sequence model which can be cast as a generalization of the Transformer model and which addresses these issues. UTs combine the parallelizability and global receptive field of feed-forward sequence models like the Transformer with the recurrent inductive bias of RNNs. We also add a dynamic per-position halting mechanism and find that it improves accuracy on several tasks. In contrast to the standard Transformer, under certain assumptions UTs can be shown to be Turing-complete. Our experiments show that UTs outperform standard Transformers on a wide range of algorithmic and language understanding tasks, including the challenging LAMBADA language modeling task where UTs achieve a new state of the art, and machine translation where UTs achieve a 0.9 BLEU improvement over Transformers on the WMT14 En-De dataset.", "target": ["Introduciamo il transformer universale, un modello per sequenze ricorrente che è parallelo nel tempo e self-attentive e che supera i transformer e le LSTM su una vasta gamma di task sequence-to-sequence, compresa la traduzione automatica.", "Propone un nuovo modello UT, basato sul modello Transformer, con aggiunta di ricorrenza e arresto dinamico della ricorrenza.", "Questo documento estende Transformer applicando ricorsivamente un blocco di multi-head self-attention, piuttosto che impilare più blocchi come nel Transformer standard."]} +{"source": "We present a framework for interpretable continual learning (ICL). We show that explanations of previously performed tasks can be used to improve performance on future tasks. ICL generates a good explanation of a finished task, then uses this to focus attention on what is important when facing a new task. The ICL idea is general and may be applied to many continual learning approaches. Here we focus on the variational continual learning framework to take advantage of its flexibility and efficacy in overcoming catastrophic forgetting. We use saliency maps to provide explanations of performed tasks and propose a new metric to assess their quality. Experiments show that ICL achieves state-of-the-art results in terms of overall continual learning performance as measured by average classification accuracy, and also in terms of its explanations, which are assessed qualitatively and quantitatively using the proposed metric.", "target": ["L'articolo sviluppa un framework di continual learning interpretabile in cui le spiegazioni dei task finiti sono utilizzate per migliorare l'attention del learner durante i task futuri, e viene proposta anche una metrica di spiegazione.", "Gli autori propongono un framework per continual learning basato su spiegazioni per classificazioni eseguite di task precedentemente appresi", "Questo articolo propone un'estensione del framework di continual learning usando il continual learning variazionale esistente come metodo di base con evidence weight."]} +{"source": "The state-of-the-art (SOTA) for mixed precision training is dominated by variants of low precision floating point operations, and in particular, FP16 accumulating into FP32 Micikevicius et al. (2017). On the other hand, while a lot of research has also happened in the domain of low and mixed-precision Integer training, these works either present results for non-SOTA networks (for instance only AlexNet for ImageNet-1K), or relatively small datasets (like CIFAR-10). In this work, we train state-of-the-art visual understanding neural networks on the ImageNet-1K dataset, with Integer operations on General Purpose (GP) hardware. In particular, we focus on Integer Fused-Multiply-and-Accumulate (FMA) operations which take two pairs of INT16 operands and accumulate results into an INT32 output.We propose a shared exponent representation of tensors and develop a Dynamic Fixed Point (DFP) scheme suitable for common neural network operations. The nuances of developing an efficient integer convolution kernel is examined, including methods to handle overflow of the INT32 accumulator. We implement CNN training for ResNet-50, GoogLeNet-v1, VGG-16 and AlexNet; and these networks achieve or exceed SOTA accuracy within the same number of iterations as their FP32 counterparts without any change in hyper-parameters and with a 1.8X improvement in end-to-end training throughput. To the best of our knowledge these results represent the first INT16 training results on GP hardware for ImageNet-1K dataset using SOTA CNNs and achieve highest reported accuracy using half precision", "target": ["Pipeline di training a precisione mista usando numeri interi a 16 bit su HW generico; accuratezza SOTA per CNN di classe ImageNet; migliore accuratezza riportata per il task di classificazione ImageNet-1K con qualsiasi training a precisione ridotta;", "Questo articolo mostra che un'attenta implementazione del calcolo dinamico in virgola fissa a precisione mista può raggiungere una precisione allo stato dell'arte utilizzando un modello di deep learning a precisione ridotta con una rappresentazione degli interi a 16 bit ", "Propone uno schema a \"virgola fissa dinamica\" che condivide la parte dell'esponente per un tensore e sviluppa procedure per fare calcoli NN con questo formato e lo dimostra per un training a precisione limitata."]} +{"source": "In this paper we introduce a new speech recognition system, leveraging a simple letter-based ConvNet acoustic model. The acoustic model requires only audio transcription for training -- no alignment annotations, nor any forced alignment step is needed. At inference, our decoder takes only a word list and a language model, and is fed with letter scores from the acoustic model -- no phonetic word lexicon is needed. Key ingredients for the acoustic model are Gated Linear Units and high dropout. We show near state-of-the-art results in word error rate on the LibriSpeech corpus with MFSC features, both on the clean and other configurations.", "target": ["Un modello acustico ConvNet basato sulle lettere porta a una pipeline di riconoscimento vocale semplice e competitiva.", "Questo articolo applica le reti neurali convoluzionali gated al riconoscimento vocale, utilizzando il criterio di training ASG."]} +{"source": "Generative adversarial networks (GANs) are a powerful framework for generative tasks. However, they are difficult to train and tend to miss modes of the true data generation process. Although GANs can learn a rich representation of the covered modes of the data in their latent space, the framework misses an inverse mapping from data to this latent space. We propose Invariant Encoding Generative Adversarial Networks (IVE-GANs), a novel GAN framework that introduces such a mapping for individual samples from the data by utilizing features in the data which are invariant to certain transformations. Since the model maps individual samples to the latent space, it naturally encourages the generator to cover all modes. We demonstrate the effectiveness of our approach in terms of generative performance and learning rich representations on several datasets including common benchmark image generation tasks.", "target": ["Un nuovo framework GAN che utilizza feature invarianti alla trasformazione per imparare rappresentazioni ricche e generatori forti.", "Propone un obiettivo GAN modificato composto da un termine GAN classico e un termine di codifica invariante.", "Questo articolo presenta l'IVE-GAN, un modello che introduce l'encoder nel framework delle Generative Adversarial Network."]} +{"source": "We propose a method for learning the dependency structure between latent variables in deep latent variable models. Our general modeling and inference framework combines the complementary strengths of deep generative models and probabilistic graphical models. In particular, we express the latent variable space of a variational autoencoder (VAE) in terms of a Bayesian network with a learned, flexible dependency structure. The network parameters, variational parameters as well as the latent topology are optimized simultaneously with a single objective. Inference is formulated via a sampling procedure that produces expectations over latent variable structures and incorporates top-down and bottom-up reasoning over latent variable values. We validate our framework in extensive experiments on MNIST, Omniglot, and CIFAR-10. Comparisons to state-of-the-art structured variational autoencoder baselines show improvements in terms of the expressiveness of the learned model.", "target": ["Proponiamo un metodo per l'apprendimento della struttura di dipendenza latente negli autoencoder variazionali.", "Utilizza una matrice di variabili casuali binarie per catturare le dipendenze tra le variabili latenti in un hierarchical deep generative model.", "Questo articolo presenta un approccio VAE in cui una struttura di dipendenza sulla variabile latente viene appresa durante il training.", "Gli autori propongono di aumentare lo spazio latente di un VAE con una struttura auto-regressiva, per migliorare l'espressività sia della rete di inferenza che del prior latente"]} +{"source": "Many real-world time series, such as in activity recognition, finance, or climate science, have changepoints where the system's structure or parameters change. Detecting changes is important as they may indicate critical events. However, existing methods for changepoint detection face challenges when (1) the patterns of change cannot be modeled using simple and predefined metrics, and (2) changes can occur gradually, at multiple time-scales. To address this, we show how changepoint detection can be treated as a supervised learning problem, and propose a new deep neural network architecture that can efficiently identify both abrupt and gradual changes at multiple scales. Our proposed method, pyramid recurrent neural network (PRNN), is designed to be scale-invariant, by incorporating wavelets and pyramid analysis techniques from multi-scale signal processing. Through experiments on synthetic and real-world datasets, we show that PRNN can detect abrupt and gradual changes with higher accuracy than the state of the art and can extrapolate to detect changepoints at novel timescales that have not been seen in training.", "target": ["Introduciamo un'architettura di rete neurale invariante alla scala per il rilevamento dei punti di cambiamento nelle serie temporali multivariate.", "Il documento sfrutta il concetto di trasformata wavelet all'interno di un'architettura di deep learning per risolvere il rilevamento dei punti di cambiamento.", "Questo articolo propone una rete neurale basata su una piramide e la applica a segnali 1D con processi sottostanti che si verificano a diverse scale temporali dove il task è la rilevazione del punto di cambiamento"]} +{"source": "We demonstrate how to learn efficient heuristics for automated reasoning algorithms through deep reinforcement learning. We focus on backtracking search algorithms for quantified Boolean logics, which already can solve formulas of impressive size - up to 100s of thousands of variables. The main challenge is to find a representation of these formulas that lends itself to making predictions in a scalable way. For challenging problems, the heuristic learned through our approach reduces execution time by a factor of 10 compared to the existing handwritten heuristics.", "target": ["Il RL trova migliori euristiche per algoritmi di ragionamento automatico.", "Mira ad apprendere un'euristica per un algoritmo di ricerca backtracking utilizzando il Reinforcement learning e propone un modello che fa uso di Graph Neural Network per produrre l'embedding di letterali e clausole, e li usa per prevedere la qualità di ogni letterale per decidere la probabilità di ogni azione.", "L'articolo propone un approccio all'apprendimento automatico di euristiche di selezione delle variabili per QBF utilizzando il deep learning"]} +{"source": "We consider the question of how to assess generative adversarial networks, in particular with respect to whether or not they generalise beyond memorising the training data. We propose a simple procedure for assessing generative adversarial network performance based on a principled consideration of what the actual goal of generalisation is. Our approach involves using a test set to estimate the Wasserstein distance between the generative distribution produced by our procedure, and the underlying data distribution. We use this procedure to assess the performance of several modern generative adversarial network architectures. We find that this procedure is sensitive to the choice of ground metric on the underlying data space, and suggest a choice of ground metric that substantially improves performance. We finally suggest that attending to the ground metric used in Wasserstein generative adversarial network training may be fruitful, and outline a concrete pathway towards doing so.", "target": ["Valutate se la vostra GAN sta effettivamente facendo qualcosa di diverso dalla memorizzazione dei dati di training.", "Mira a fornire una misura/test di qualità per le GAN e propone di valutare l'approssimazione attuale di una distribuzione appresa da una GAN utilizzando la distanza di Wasserstein tra due distribuzioni composte da una somma di Dirac come baseline di performance.", "Questo articolo ha proposto una procedura per valutare le prestazioni delle GAN riconsiderando la chiave di osservazione, utilizzando la procedura per testare e migliorare le GAN attuali"]} +{"source": "The choice of activation functions in deep networks has a significant effect on the training dynamics and task performance. Currently, the most successful and widely-used activation function is the Rectified Linear Unit (ReLU). Although various hand-designed alternatives to ReLU have been proposed, none have managed to replace it due to inconsistent gains. In this work, we propose to leverage automatic search techniques to discover new activation functions. Using a combination of exhaustive and reinforcement learning-based search, we discover multiple novel activation functions. We verify the effectiveness of the searches by conducting an empirical evaluation with the best discovered activation function. Our experiments show that the best discovered activation function, f(x) = x * sigmoid(beta * x), which we name Swish, tends to work better than ReLU on deeper models across a number of challenging datasets. For example, simply replacing ReLUs with Swish units improves top-1 classification accuracy on ImageNet by 0.9% for Mobile NASNet-A and 0.6% for Inception-ResNet-v2. The simplicity of Swish and its similarity to ReLU make it easy for practitioners to replace ReLUs with Swish units in any neural network.", "target": ["Usiamo tecniche di ricerca per scoprire nuove funzioni di attivazione, e la nostra migliore funzione di attivazione scoperta, f(x) = x * sigmoid(beta * x), supera ReLU su una serie di task impegnativi come ImageNet.", "Propone un approccio basato sul reinforcement learning per trovare le non linearità cercando attraverso combinazioni da un insieme di operatori unari e binari.", "Questo articolo utilizza reinforcement learning per cercare la combinazione di un insieme di funzioni unarie e binarie che risultino in una nuova funzione di attivazione", "L'autore usa il reinforcement learning per trovare nuove potenziali funzioni di attivazione da un ricco insieme di possibili candidati."]} +{"source": "Successful training of convolutional neural networks is often associated with suffi- ciently deep architectures composed of high amounts of features. These networks typically rely on a variety of regularization and pruning techniques to converge to less redundant states. We introduce a novel bottom-up approach to expand representations in fixed-depth architectures. These architectures start from just a single feature per layer and greedily increase width of individual layers to attain effective representational capacities needed for a specific task. While network growth can rely on a family of metrics, we propose a computationally efficient version based on feature time evolution and demonstrate its potency in determin- ing feature importance and a networks’ effective capacity. We demonstrate how automatically expanded architectures converge to similar topologies that benefit from lesser amount of parameters or improved accuracy and exhibit systematic correspondence in representational complexity with the specified task. In contrast to conventional design patterns with a typical monotonic increase in the amount of features with increased depth, we observe that CNNs perform better when there is more learnable parameters in intermediate, with falloffs to earlier and later layers.", "target": ["Un algoritmo bottom-up che espande le CNN a partire da una feature per layer fino ad architetture con sufficiente capacità di rappresentazione.", "Propone di regolare dinamicamente la profondità della feature map di una rete neurale completamente convoluzionale, formulando una misura di autosomiglianza e aumentando le prestazioni.", "Introduce una semplice metrica basata sulla correlazione per misurare se i filtri nelle reti neurali sono usati efficacemente, come proxy della capacità effettiva.", "Mira ad affrontare il problema della ricerca dell'architettura di deep learning attraverso l'aggiunta e la rimozione incrementale di canali nei layer intermedi della rete."]} +{"source": "Deep neural networks are almost universally trained with reverse-mode automatic differentiation (a.k.a. backpropagation). Biological networks, on the other hand, appear to lack any mechanism for sending gradients back to their input neurons, and thus cannot be learning in this way. In response to this, Scellier & Bengio (2017) proposed Equilibrium Propagation - a method for gradient-based train- ing of neural networks which uses only local learning rules and, crucially, does not rely on neurons having a mechanism for back-propagating an error gradient. Equilibrium propagation, however, has a major practical limitation: inference involves doing an iterative optimization of neural activations to find a fixed-point, and the number of steps required to closely approximate this fixed point scales poorly with the depth of the network. In response to this problem, we propose Initialized Equilibrium Propagation, which trains a feedforward network to initialize the iterative inference procedure for Equilibrium propagation. This feed-forward network learns to approximate the state of the fixed-point using a local learning rule. After training, we can simply use this initializing network for inference, resulting in a learned feedforward network. Our experiments show that this network appears to work as well or better than the original version of Equilibrium propagation. This shows how we might go about training deep networks without using backpropagation.", "target": ["Addestriamo una rete feedforward senza backprop utilizzando un modello basato sull'energia per fornire obiettivi locali", "Questo articolo mira a velocizzare la procedura di inferenza iterativa nei modelli basati sull'energia addestrati con Equilibrium Propagation (EP), proponendo di addestrare una rete feedforward per prevedere un punto fisso della \"rete equilibrante\". ", "Addestrare una rete separata per inizializzare le reti ricorrenti addestrate utilizzando la propagazione dell'equilibrio"]} +{"source": "We propose a novel generative model architecture designed to learn representations for images that factor out a single attribute from the rest of the representation. A single object may have many attributes which when altered do not change the identity of the object itself. Consider the human face; the identity of a particular person is independent of whether or not they happen to be wearing glasses. The attribute of wearing glasses can be changed without changing the identity of the person. However, the ability to manipulate and alter image attributes without altering the object identity is not a trivial task. Here, we are interested in learning a representation of the image that separates the identity of an object (such as a human face) from an attribute (such as 'wearing glasses'). We demonstrate the success of our factorization approach by using the learned representation to synthesize the same face with and without a chosen attribute. We refer to this specific synthesis process as image attribute manipulation. We further demonstrate that our model achieves competitive scores, with state of the art, on a facial attribute classification task.", "target": ["Imparare le rappresentazioni per le immagini che possono fattorizzare un singolo attributo.", "Questo paper si basa su GAN con VAE condizionali per consentire la manipolazione degli attributi nel processo di sintesi.", "Questo articolo propone un modello generativo per imparare la rappresentazione che può separare l'identità di un oggetto da un attributo, ed estende l'adversarial autoencoder aggiungendo una rete ausiliaria."]} +{"source": "Stochastic video prediction models take in a sequence of image frames, and generate a sequence of consecutive future image frames. These models typically generate future frames in an autoregressive fashion, which is slow and requires the input and output frames to be consecutive. We introduce a model that overcomes these drawbacks by generating a latent representation from an arbitrary set of frames that can then be used to simultaneously and efficiently sample temporally consistent frames at arbitrary time-points. For example, our model can \"jump\" and directly sample frames at the end of the video, without sampling intermediate frames. Synthetic video evaluations confirm substantial gains in speed and functionality without loss in fidelity. We also apply our framework to a 3D scene reconstruction dataset. Here, our model is conditioned on camera location and can sample consistent sets of images for what an occluded region of a 3D scene might look like, even if there are multiple possibilities for what that region might contain. Reconstructions and videos are available at https://bit.ly/2O4Pc4R.", "target": ["Presentiamo un modello per la ricostruzione 3D coerente e la jumpy video prediction, ad esempio la produzione di frame di immagini a più passi temporali nel futuro senza generare frame intermedi.", "Questo articolo propone un metodo generale per la modellazione dei dati indicizzati codificando le informazioni dell'indice insieme all'osservazione in una rete neurale, e poi decodificare la condizione di osservazione sull'indice di destinazione.", "Propone di utilizzare un VAE che codifica il video in input in modo invariante alla permutazione per prevedere i futuri frame di un video."]} +{"source": "The ADAM optimizer is exceedingly popular in the deep learning community. Often it works very well, sometimes it doesn’t. Why? We interpret ADAM as a combination of two aspects: for each weight, the update direction is determined by the sign of the stochastic gradient, whereas the update magnitude is solely determined by an estimate of its relative variance. We disentangle these two aspects and analyze them in isolation, shedding light on ADAM ’s inner workings. Transferring the \"variance adaptation” to momentum- SGD gives rise to a novel method, completing the practitioner’s toolbox for problems where ADAM fails.", "target": ["Si analizza il popolare ottimizzatore Adam", "L'articolo cerca di migliorare Adam basandosi sull'adattamento della varianza con momentum proponendo due algoritmi", "Questo articolo analizza l'invarianza di scala e la forma particolare del learning rate utilizzato in Adam, sostenendo che l'aggiornamento di Adam è una combinazione di un sign-update e un learning rate basato sulla varianza.", "L'articolo divide l'algoritmo ADAM in due componenti: direzione stocastica nel segno del gradiente e step size adattiva con varianza relativa, e due algoritmi vengono proposti per testare ciascuno di essi."]} +{"source": "We propose a novel framework to adaptively adjust the dropout rates for the deep neural network based on a Rademacher complexity bound. The state-of-the-art deep learning algorithms impose dropout strategy to prevent feature co-adaptation. However, choosing the dropout rates remains an art of heuristics or relies on empirical grid-search over some hyperparameter space. In this work, we show the network Rademacher complexity is bounded by a function related to the dropout rate vectors and the weight coefficient matrices. Subsequently, we impose this bound as a regularizer and provide a theoretical justified way to trade-off between model complexity and representation power. Therefore, the dropout rates and the empirical loss are unified into the same objective function, which is then optimized using the block coordinate descent algorithm. We discover that the adaptively adjusted dropout rates converge to some interesting distributions that reveal meaningful patterns.Experiments on the task of image and document classification also show our method achieves better performance compared to the state-of the-art dropout algorithms.", "target": ["Proponiamo un nuovo framework per regolare in modo adattivo i dropout rate per le deep neural network basato su un bound di complessità di Rademacher.", "Gli autori collegano i parametri di dropout a un bound della complessità di Rademacher della rete", "Mette in relazione la complessità dell'apprendibilità delle reti con i dropout rate nella backpropagation."]} +{"source": "Sensor fusion is a key technology that integrates various sensory inputs to allow for robust decision making in many applications such as autonomous driving and robot control. Deep neural networks have been adopted for sensor fusion in a body of recent studies. Among these, the so-called netgated architecture was proposed, which has demonstrated improved performances over the conventional convolu- tional neural networks (CNN). In this paper, we address several limitations of the baseline negated architecture by proposing two further optimized architectures: a coarser-grained gated architecture employing (feature) group-level fusion weights and a two-stage gated architectures leveraging both the group-level and feature- level fusion weights. Using driving mode prediction and human activity recogni- tion datasets, we demonstrate the significant performance improvements brought by the proposed gated architectures and also their robustness in the presence of sensor noise and failures.", "target": ["Vengono proposte architetture di deep learning gated ottimizzate per la fusione dei sensori.", "Gli autori migliorano diverse limitazioni della baseline dell'architettura negata proponendo un'architettura di fusione gated più grossolana e un'architettura di fusione gated a due stadi", "Propone due architetture di deep learning gated per la fusione dei sensori e, avendo le feature raggruppate, dimostra prestazioni migliori, soprattutto in presenza di rumore casuale dei sensori e di guasti."]} +{"source": "We develop a mean field theory for batch normalization in fully-connected feedforward neural networks. In so doing, we provide a precise characterization of signal propagation and gradient backpropagation in wide batch-normalized networks at initialization. Our theory shows that gradient signals grow exponentially in depth and that these exploding gradients cannot be eliminated by tuning the initial weight variances or by adjusting the nonlinear activation function. Indeed, batch normalization itself is the cause of gradient explosion. As a result, vanilla batch-normalized networks without skip connections are not trainable at large depths for common initialization schemes, a prediction that we verify with a variety of empirical simulations. While gradient explosion cannot be eliminated, it can be reduced by tuning the network close to the linear regime, which improves the trainability of deep batch-normalized networks without residual connections. Finally, we investigate the learning dynamics of batch-normalized networks and observe that after a single step of optimization the networks achieve a relatively stable equilibrium in which gradients have dramatically smaller dynamic range. Our theory leverages Laplace, Fourier, and Gegenbauer transforms and we derive new identities that may be of independent interest.", "target": ["La batch normalization causa l'esplosione dei gradienti nelle reti feedforward standard.", "Sviluppa una teoria del campo medio per la batch normalization (BN) in reti completamente connesse con pesi inizializzati in modo casuale.", "Fornisce una prospettiva dinamica sulle deep neural network utilizzando l'evoluzione della matrice di covarianza insieme ai layer."]} +{"source": "We present NeuroSAT, a message passing neural network that learns to solve SAT problems after only being trained as a classifier to predict satisfiability. Although it is not competitive with state-of-the-art SAT solvers, NeuroSAT can solve problems that are substantially larger and more difficult than it ever saw during training by simply running for more iterations. Moreover, NeuroSAT generalizes to novel distributions; after training only on random SAT problems, at test time it can solve SAT problems encoding graph coloring, clique detection, dominating set, and vertex cover problems, all on a range of distributions over small random graphs.", "target": ["Addestriamo una graph network per prevedere la soddisfacibilità booleana e mostriamo che impara a cercare soluzioni, e che le soluzioni che trova possono essere decodificate dalle sue attivazioni.", "L'articolo descrive un'architettura generale di rete neurale per prevedere la soddisfacibilità", "Questo articolo presenta l'architettura NeuroSAT che utilizza una deep neural network con message passing per prevedere la soddisfacibilità delle istanze CNF"]} +{"source": "Spatiotemporal forecasting has various applications in neuroscience, climate and transportation domain. Traffic forecasting is one canonical example of such learning task. The task is challenging due to (1) complex spatial dependency on road networks, (2) non-linear temporal dynamics with changing road conditions and (3) inherent difficulty of long-term forecasting. To address these challenges, we propose to model the traffic flow as a diffusion process on a directed graph and introduce Diffusion Convolutional Recurrent Neural Network (DCRNN), a deep learning framework for traffic forecasting that incorporates both spatial and temporal dependency in the traffic flow. Specifically, DCRNN captures the spatial dependency using bidirectional random walks on the graph, and the temporal dependency using the encoder-decoder architecture with scheduled sampling. We evaluate the framework on two real-world large-scale road network traffic datasets and observe consistent improvement of 12% - 15% over state-of-the-art baselines", "target": ["Un modello neurale per sequenze che impara a prevedere su un grafo diretto.", "L'articolo propone l'architettura Diffusion Convolutional Recurrent Neural Network per il problema della predizione del traffico spazio-temporale", "Propone di costruire un modello di predizione del traffico usando un processo di diffusione per reti neurali ricorrenti convoluzionali per affrontare l'autocorrelazione spazio-temporale."]} +{"source": "Obtaining reliable uncertainty estimates of neural network predictions is a long standing challenge. Bayesian neural networks have been proposed as a solution, but it remains open how to specify their prior. In particular, the common practice of a standard normal prior in weight space imposes only weak regularities, causing the function posterior to possibly generalize in unforeseen ways on inputs outside of the training distribution. We propose noise contrastive priors (NCPs) to obtain reliable uncertainty estimates. The key idea is to train the model to output high uncertainty for data points outside of the training distribution. NCPs do so using an input prior, which adds noise to the inputs of the current mini batch, and an output prior, which is a wide distribution given these inputs. NCPs are compatible with any model that can output uncertainty estimates, are easy to scale, and yield reliable uncertainty estimates throughout training. Empirically, we show that NCPs prevent overfitting outside of the training distribution and result in uncertainty estimates that are useful for active learning. We demonstrate the scalability of our method on the flight delays data set, where we significantly improve upon previously published results.", "target": ["Addestriamo le reti neurali ad essere incerte su input rumorosi per evitare predizioni troppo sicure al di fuori della distribuzione di training.", "Presenta un approccio per ottenere stime di incertezza per le predizioni delle reti neurali che ha buone prestazioni quando si quantifica l'incertezza predittiva in punti che sono al di fuori della distribuzione di training.", "L'articolo considera il problema della stima dell'incertezza delle reti neurali e propone di utilizzare l'approccio bayesiano con prior noise contrastive"]} +{"source": "Convolutional neural networks (CNNs) were inspired by human vision and, in some settings, achieve a performance comparable to human object recognition. This has lead to the speculation that both systems use similar mechanisms to perform recognition. In this study, we conducted a series of simulations that indicate that there is a fundamental difference between human vision and CNNs: while object recognition in humans relies on analysing shape, CNNs do not have such a shape-bias. We teased apart the type of features selected by the model by modifying the CIFAR-10 dataset so that, in addition to containing objects with shape, the images concurrently contained non-shape features, such as a noise-like mask. When trained on these modified set of images, the model did not show any bias towards selecting shapes as features. Instead it relied on whichever feature allowed it to perform the best prediction -- even when this feature was a noise-like mask or a single predictive pixel amongst 50176 pixels. We also found that regularisation methods, such as batch normalisation or Dropout, did not change this behaviour and neither did past or concurrent experience with images from other datasets.", "target": ["Questo studio mette in evidenza una differenza chiave tra la visione umana e le CNN: mentre il riconoscimento degli oggetti negli esseri umani si basa sull'analisi della forma, le CNN non hanno una tale bias per la forma.", "Cerca di stabilire attraverso una serie di esperimenti ben progettati che le CNN addestrate per la classificazione delle immagini non codificano i bias di forma come nella visione umana.", "Questo articolo mette in evidenza il fatto che le CNN non imparano necessariamente a riconoscere gli oggetti in base alla loro forma e mostra che si fidano troppo delle feature basate sul rumore."]} +{"source": "The development of high-dimensional generative models has recently gained a great surge of interest with the introduction of variational auto-encoders and generative adversarial neural networks. Different variants have been proposed where the underlying latent space is structured, for example, based on attributes describing the data to generate. We focus on a particular problem where one aims at generating samples corresponding to a number of objects under various views. We assume that the distribution of the data is driven by two independent latent factors: the content, which represents the intrinsic features of an object, and the view, which stands for the settings of a particular observation of that object. Therefore, we propose a generative model and a conditional variant built on such a disentangled latent space. This approach allows us to generate realistic samples corresponding to various objects in a high variety of views. Unlike many multi-view approaches, our model doesn't need any supervision on the views but only on the content. Compared to other conditional generation approaches that are mostly based on binary or categorical attributes, we make no such assumption about the factors of variations. Our model can be used on problems with a huge, potentially infinite, number of categories. We experiment it on four images datasets on which we demonstrate the effectiveness of the model and its ability to generalize.", "target": ["Descriviamo un nuovo modello generativo multi-view che può generare più view dello stesso oggetto, o più oggetti nella stessa view senza bisogno di label sulle view.", "Questo articolo presenta un metodo basato su GAN per la generazione di immagini che cerca di separare le variabili latenti che descrivono il contenuto dell'immagine da quelle che descrivono le proprietà della view.", "Questo articolo propone un'architettura GAN che mira a decomporre la distribuzione sottostante di una particolare classe in \"contenuto\" e \"view\".", "Propone un nuovo modello generativo basato sulla Generative Adversarial Network (GAN) che separa il contenuto e la view degli oggetti senza supervisione della view ed estende GMV in un modello generativo condizionale che prende un'immagine di input e genera diverse view dell'oggetto nell'immagine di input."]} +{"source": "The huge size of deep networks hinders their use in small computing devices. In this paper, we consider compressing the network by weight quantization. We extend a recently proposed loss-aware weight binarization scheme to ternarization, with possibly different scaling parameters for the positive and negative weights, and m-bit (where m > 2) quantization. Experiments on feedforward and recurrent neural networks show that the proposed scheme outperforms state-of-the-art weight quantization algorithms, and is as accurate (or even more accurate) than the full-precision network.", "target": ["Viene proposto un algoritmo di quantizzazione del peso consapevole della loss che considera direttamente il suo effetto sulla loss.", "Propone un metodo di compressione della rete per mezzo della ternarizzazione dei pesi.", "L'articolo propone un nuovo metodo per addestrare DNN con pesi quantizzati, includendo la quantizzazione come vincolo in un algoritmo quasi-Newton prossimale, che apprende simultaneamente uno scaling per i valori quantizzati.", "L'articolo estende lo schema di binarizzazione del peso consapevole della loss alla ternarizzazione e alla quantizzazione arbitraria a m-bit e dimostra le sue promettenti prestazioni."]} +{"source": "In the pursuit of increasingly intelligent learning systems, abstraction plays a vital role in enabling sophisticated decisions to be made in complex environments. The options framework provides formalism for such abstraction over sequences of decisions. However most models require that options be given a priori, presumably specified by hand, which is neither efficient, nor scalable. Indeed, it is preferable to learn options directly from interaction with the environment. Despite several efforts, this remains a difficult problem: many approaches require access to a model of the environmental dynamics, and inferred options are often not interpretable, which limits our ability to explain the system behavior for verification or debugging purposes. In this work we develop a novel policy gradient method for the automatic learning of policies with options. This algorithm uses inference methods to simultaneously improve all of the options available to an agent, and thus can be employed in an off-policy manner, without observing option labels. Experimental results show that the options learned can be interpreted. Further, we find that the method presented here is more sample efficient than existing methods, leading to faster and more stable learning of policies with options.", "target": ["Sviluppiamo un nuovo metodo di policy gradient per l'apprendimento automatico di policy con opzioni utilizzando una fase di inferenza differenziabile.", "L'articolo presenta una nuova tecnica di policy gradient per l'apprendimento delle opzioni, dove un singolo sample può essere usato per aggiornare tutte le opzioni.", "Propone un metodo off-policy per l'apprendimento di opzioni in problemi continui complessi."]} +{"source": "The paper, interested in unsupervised feature selection, aims to retain the features best accounting for the local patterns in the data. The proposed approach, called Locally Linear Unsupervised Feature Selection, relies on a dimensionality reduction method to characterize such patterns; each feature is thereafter assessed according to its compliance w.r.t. the local patterns, taking inspiration from Locally Linear Embedding (Roweis and Saul, 2000). The experimental validation of the approach on the scikit-feature benchmark suite demonstrates its effectiveness compared to the state of the art.", "target": ["Selezione non supervisionata delle feature attraverso la cattura della struttura lineare e locale dei dati", "Propone una selezione delle feature non supervisionata localmente lineare.", "L'articolo propone il metodo LLUFS per la selezione delle feature."]} +{"source": "Humans can understand and produce new utterances effortlessly, thanks to their systematic compositional skills. Once a person learns the meaning of a new verb \"dax,\" he or she can immediately understand the meaning of \"dax twice\" or \"sing and dax.\" In this paper, we introduce the SCAN domain, consisting of a set of simple compositional navigation commands paired with the corresponding action sequences. We then test the zero-shot generalization capabilities of a variety of recurrent neural networks (RNNs) trained on SCAN with sequence-to-sequence methods. We find that RNNs can generalize well when the differences between training and test commands are small, so that they can apply \"mix-and-match\" strategies to solve the task. However, when generalization requires systematic compositional skills (as in the \"dax\" example above), RNNs fail spectacularly. We conclude with a proof-of-concept experiment in neural machine translation, supporting the conjecture that lack of systematicity is an important factor explaining why neural networks need very large training sets.", "target": ["Utilizzando un semplice task di navigazione guidato dal linguaggio, studiamo le capacità compositive delle moderne reti ricorrenti seq2seq.", "Questo articolo si concentra sulle capacità compositive di zero-shot learning delle moderne RNN sequence-to-sequence ed espone i difetti delle attuali architetture RNN seq2seq.", "L'articolo analizza le capacità di composizione delle RNN, in particolare, la capacità di generalizzazione delle RNN su un sottoinsieme casuale di comandi SCAN, su comandi SCAN più lunghi, e sulla composizione di comandi primitivi.", "Gli autori introducono un nuovo dataset che facilita l'analisi di un caso di apprendimento Seq2Seq"]} +{"source": "This paper addresses the challenging problem of retrieval and matching of graph structured objects, and makes two key contributions. First, we demonstrate how Graph Neural Networks (GNN), which have emerged as an effective model for various supervised prediction problems defined on structured data, can be trained to produce embedding of graphs in vector spaces that enables efficient similarity reasoning. Second, we propose a novel Graph Matching Network model that, given a pair of graphs as input, computes a similarity score between them by jointly reasoning on the pair through a new cross-graph attention-based matching mechanism. We demonstrate the effectiveness of our models on different domains including the challenging problem of control-flow-graph based function similarity search that plays an important role in the detection of vulnerabilities in software systems. The experimental analysis demonstrates that our models are not only able to exploit structure in the context of similarity learning but they can also outperform domain-specific baseline systems that have been carefully hand-engineered for these problems.", "target": ["Affrontiamo il problema del similarity learning per oggetti strutturati con applicazioni in particolare nella sicurezza informatica, e proponiamo un nuovo modello di graph matching network che eccelle in questo task.", "Gli autori introducono una Graph Matching Network per il recupero e la corrispondenza di oggetti strutturati a grafo.", "Gli autori attaccano il problema del matching dei grafi proponendo un'estensione delle reti di embedding dei grafi", "Gli autori presentano due metodi per l'apprendimento di un punteggio di somiglianza tra coppie di grafi e mostrano l'utilità di introdurre idee dal graph matching alle graph neural network."]} +{"source": "Context information plays an important role in human language understanding, and it is also useful for machines to learn vector representations of language. In this paper, we explore an asymmetric encoder-decoder structure for unsupervised context-based sentence representation learning. As a result, we build an encoder-decoder architecture with an RNN encoder and a CNN decoder, and we show that neither an autoregressive decoder nor an RNN decoder is required. We further combine a suite of effective designs to significantly improve model efficiency while also achieving better performance. Our model is trained on two different large unlabeled corpora, and in both cases transferability is evaluated on a set of downstream language understanding tasks. We empirically show that our model is simple and fast while producing rich sentence representations that excel in downstream tasks.", "target": ["Abbiamo proposto un modello RNN-CNN encoder-decoder per un veloce apprendimento unsupervised di sentence embedding.", "Modifiche al framework di skip-thought per l'apprendimento di embedding di frasi.", "Questo articolo presenta un nuovo design ibrido costituito da un encoder RNN e un decoder CNN per l'uso nel pretraining, che non richiede un decoder autoregressivo quando si fa pretraining degli encoder.", "Gli autori estendono Skip-thought decodificando solo una frase target utilizzando un decoder CNN."]} +{"source": "Building on the success of deep learning, two modern approaches to learn a probability model of the observed data are Generative Adversarial Networks (GANs) and Variational AutoEncoders (VAEs). VAEs consider an explicit probability model for the data and compute a generative distribution by maximizing a variational lower-bound on the log-likelihood function. GANs, however, compute a generative model by minimizing a distance between observed and generated probability distributions without considering an explicit model for the observed data. The lack of having explicit probability models in GANs prohibits computation of sample likelihoods in their frameworks and limits their use in statistical inference problems. In this work, we show that an optimal transport GAN with the entropy regularization can be viewed as a generative model that maximizes a lower-bound on average sample likelihoods, an approach that VAEs are based on. In particular, our proof constructs an explicit probability model for GANs that can be used to compute likelihood statistics within GAN's framework. Our numerical results on several datasets demonstrate consistent trends with the proposed theory.", "target": ["Un approccio statistico per calcolare le verosimiglianze del sample nelle generative adversarial network", "Mostrare che WGAN con regolarizzazione entropica massimizza un lower bound sulla probabilità della distribuzione dei dati osservati.", "Gli autori sostengono che è possibile sfruttare l'upper bound di un trasporto ottimale regolarizzato dall'entropia per ottenere una misura di \"likelihood del sample\"."]} +{"source": "We introduce geomstats, a Python package for Riemannian modelization and optimization over manifolds such as hyperspheres, hyperbolic spaces, SPD matrices or Lie groups of transformations. Our contribution is threefold. First, geomstats allows the flexible modeling of many a machine learning problem through an efficient and extensively unit-tested implementations of these manifolds, as well as the set of useful Riemannian metrics, exponential and logarithm maps that we provide. Moreover, the wide choice of loss functions and our implementation of the corresponding gradients allow fast and easy optimization over manifolds. Finally, geomstats is the only package to provide a unified framework for Riemannian geometry, as the operations implemented in geomstats are available with different computing backends (numpy,tensorflow and keras), as well as with a GPU-enabled mode–-thus considerably facilitating the application of Riemannian geometry in machine learning. In this paper, we present geomstats through a review of the utility and advantages of manifolds in machine learning, using the concrete examples that they span to show the efficiency and practicality of their implementation using our package", "target": ["Introduciamo geomstats, un efficiente pacchetto Python per la modellazione e l'ottimizzazione Riemanniana su manifold compatibile sia con numpy che con tensorflow.", "L'articolo introduce il pacchetto software geomstats, che fornisce un uso semplice dei manifold e delle metriche Riemanniane all'interno dei modelli di apprendimento automatico", "Propone un pacchetto Python per l'ottimizzazione e le applicazioni su manifold Riemanniani ed evidenzia le differenze tra il pacchetto Geomstats e altri pacchetti.", "Introduce un toolbox geometrico, Geomstats, per machine learning su manifold Riemanniani."]} +{"source": "We propose to execute deep neural networks (DNNs) with dynamic and sparse graph (DSG) structure for compressive memory and accelerative execution during both training and inference. The great success of DNNs motivates the pursuing of lightweight models for the deployment onto embedded devices. However, most of the previous studies optimize for inference while neglect training or even complicate it. Training is far more intractable, since (i) the neurons dominate the memory cost rather than the weights in inference; (ii) the dynamic activation makes previous sparse acceleration via one-off optimization on fixed weight invalid; (iii) batch normalization (BN) is critical for maintaining accuracy while its activation reorganization damages the sparsity. To address these issues, DSG activates only a small amount of neurons with high selectivity at each iteration via a dimensionreduction search and obtains the BN compatibility via a double-mask selection. Experiments show significant memory saving (1.7-4.5x) and operation reduction (2.3-4.4x) with little accuracy loss on various benchmarks.", "target": ["Costruiamo il grafo dinamico sparso tramite la ricerca di dimension-reduction per ridurre i costi di calcolo e di memoria sia nel training di DNN che nell'inferenza.", "Gli autori propongono di utilizzare il grafo computazionale dinamico e sparso per ridurre la memoria di calcolo e il costo temporale nelle deep neural network (DNN).", "Questo articolo propone un metodo per accelerare il training e l'inferenza delle deep neural network usando il pruning dinamico del grafo computazionale."]} +{"source": "Efficient exploration remains a major challenge for reinforcement learning. One reason is that the variability of the returns often depends on the current state and action, and is therefore heteroscedastic. Classical exploration strategies such as upper confidence bound algorithms and Thompson sampling fail to appropriately account for heteroscedasticity, even in the bandit setting. Motivated by recent findings that address this issue in bandits, we propose to use Information-Directed Sampling (IDS) for exploration in reinforcement learning. As our main contribution, we build on recent advances in distributional reinforcement learning and propose a novel, tractable approximation of IDS for deep Q-learning. The resulting exploration strategy explicitly accounts for both parametric uncertainty and heteroscedastic observation noise. We evaluate our method on Atari games and demonstrate a significant improvement over alternative approaches.", "target": ["Sviluppiamo un'estensione pratica dell'Information-Directed Sampling per Reinforcement Learning, che tiene conto dell'incertezza parametrica e dell'eteroscedasticità nella distribuzione di ritorno per l'esplorazione.", "Gli autori propongono un modo di estendere l'Information-Directed Sampling al reinforcement learning combinando due tipi di incertezza per ottenere una semplice strategia di esplorazione basata su IDS.", "Questo articolo studia approcci sofisticati di esplorazione per reinforcement learning costruiti su Information Direct Sampling e su Distributional Reinforcement Learning"]} +{"source": "We address the problem of learning structured policies for continuous control. In traditional reinforcement learning, policies of agents are learned by MLPs which take the concatenation of all observations from the environment as input for predicting actions. In this work, we propose NerveNet to explicitly model the structure of an agent, which naturally takes the form of a graph. Specifically, serving as the agent's policy network, NerveNet first propagates information over the structure of the agent and then predict actions for different parts of the agent. In the experiments, we first show that our NerveNet is comparable to state-of-the-art methods on standard MuJoCo environments. We further propose our customized reinforcement learning environments for benchmarking two types of structure transfer learning tasks, i.e., size and disability transfer. We demonstrate that policies learned by NerveNet are significantly better than policies learned by other models and are able to transfer even in a zero-shot setting.", "target": ["Si utilizzano le graph neural network per modellare le informazioni strutturali degli agenti per migliorare la policy e la trasferibilità", "Un metodo per la rappresentazione e l'apprendimento di una policy strutturata per task di controllo continuo utilizzando Graph Neural Network", "Il paper propone l'aggiunta di un ulteriore struttura nei problemi di reinforcement learning, in particolare la struttura della morfologia dell'agente", "Si propone un'applicazione delle Graph Neural Network all'apprendimento di policy per il controllo di robot \"centipede\" di diverse lunghezze."]} +{"source": "Real-world tasks are often highly structured. Hierarchical reinforcement learning (HRL) has attracted research interest as an approach for leveraging the hierarchical structure of a given task in reinforcement learning (RL). However, identifying the hierarchical policy structure that enhances the performance of RL is not a trivial task. In this paper, we propose an HRL method that learns a latent variable of a hierarchical policy using mutual information maximization. Our approach can be interpreted as a way to learn a discrete and latent representation of the state-action space. To learn option policies that correspond to modes of the advantage function, we introduce advantage-weighted importance sampling. In our HRL method, the gating policy learns to select option policies based on an option-value function, and these option policies are optimized based on the deterministic policy gradient method. This framework is derived by leveraging the analogy between a monolithic policy in standard RL and a hierarchical policy in HRL by using a deterministic option policy. Experimental results indicate that our HRL approach can learn a diversity of options and that it can enhance the performance of RL in continuous control tasks.", "target": ["Questo articolo presenta un framework di reinforcement learning gerarchico basato su policy di opzioni deterministiche e massimizzazione dell'informazione mutua.", "Propone un algoritmo HRL che cerca di imparare le opzioni che massimizzano la loro mutua informazione con la densità di stato-azione sotto la policy ottimale.", "Questo articolo propone un sistema HRL in cui l'informazione mutua della variabile latente e delle coppie stato-azione è approssimativamente massimizzata.", "Propone un criterio che mira a massimizzare l'informazione mutua tra le opzioni e le coppie stato-azione e mostra empiricamente che le opzioni apprese decompongono lo spazio stato-azione ma non lo spazio di stato."]} +{"source": "Deep neural networks (DNNs) have achieved impressive predictive performance due to their ability to learn complex, non-linear relationships between variables. However, the inability to effectively visualize these relationships has led to DNNs being characterized as black boxes and consequently limited their applications. To ameliorate this problem, we introduce the use of hierarchical interpretations to explain DNN predictions through our proposed method: agglomerative contextual decomposition (ACD). Given a prediction from a trained DNN, ACD produces a hierarchical clustering of the input features, along with the contribution of each cluster to the final prediction. This hierarchy is optimized to identify clusters of features that the DNN learned are predictive. We introduce ACD using examples from Stanford Sentiment Treebank and ImageNet, in order to diagnose incorrect predictions, identify dataset bias, and extract polarizing phrases of varying lengths. Through human experiments, we demonstrate that ACD enables users both to identify the more accurate of two DNNs and to better trust a DNN's outputs. We also find that ACD's hierarchy is largely robust to adversarial perturbations, implying that it captures fundamental aspects of the input and ignores spurious noise.", "target": ["Introduciamo e convalidiamo le interpretazioni locali gerarchiche, la prima tecnica per cercare e visualizzare automaticamente le interazioni importanti per le predizioni individuali fatte da LSTM e CNN.", "Un nuovo approccio per spiegare le predizioni delle reti neurali imparando rappresentazioni gerarchiche di gruppi di feature di input e il loro contributo alla predizione finale", "Estende un metodo di interpretazione delle feature esistente per LSTM alle DNN più generiche e introduce un clustering gerarchico delle feature di input e i contributi di ogni cluster alla predizione finale.", "Questo articolo propone un'estensione gerarchica della decomposizione contestuale."]} +{"source": "Principal Filter Analysis (PFA) is an easy to implement, yet effective method for neural network compression. PFA exploits the intrinsic correlation between filter responses within network layers to recommend a smaller network footprint. We propose two compression algorithms: the first allows a user to specify the proportion of the original spectral energy that should be preserved in each layer after compression, while the second is a heuristic that leads to a parameter-free approach that automatically selects the compression used at each layer. Both algorithms are evaluated against several architectures and datasets, and we show considerable compression rates without compromising accuracy, e.g., for VGG-16 on CIFAR-10, CIFAR-100 and ImageNet, PFA achieves a compression rate of 8x, 3x, and 1.4x with an accuracy gain of 0.4%, 1.4% points, and 2.4% respectively. In our tests we also demonstrate that networks compressed with PFA achieve an accuracy that is very close to the empirical upper bound for a given compression ratio. Finally, we show how PFA is an effective tool for simultaneous compression and domain adaptation.", "target": ["Proponiamo un metodo facile da implementare ma efficace per la compressione delle reti neurali. PFA sfrutta la correlazione intrinseca tra le risposte dei filtri all'interno dei layer della rete per fornire una rete più piccola.", "Propone di fare pruning delle reti convoluzionali analizzando la correlazione osservata tra i filtri di uno stesso layer come espresso dallo spettro di autovalori della loro matrice di covarianza.", "Questo articolo introduce un approccio alla compressione delle reti neurali guardando la correlazione delle risposte dei filtri in ogni layer attraverso due strategie.", "Questo documento propone un metodo di compressione basato sull'analisi spettrale"]} +{"source": "We propose a method to efficiently learn diverse strategies in reinforcement learning for query reformulation in the tasks of document retrieval and question answering. In the proposed framework an agent consists of multiple specialized sub-agents and a meta-agent that learns to aggregate the answers from sub-agents to produce a final answer. Sub-agents are trained on disjoint partitions of the training data, while the meta-agent is trained on the full training set. Our method makes learning faster, because it is highly parallelizable, and has better generalization performance than strong baselines, such as an ensemble of agents trained on the full data. We show that the improved performance is due to the increased diversity of reformulation strategies.", "target": ["Agenti di riformulazione di query multiple e diversificate addestrati con reinforcement learning per migliorare i motori di ricerca.", "Parellelizzazione del metodo dell'ensemble nel reinforcement learning per la riformulazione delle query, accelerando il training e migliorando la diversità delle riformulazioni apprese", "Gli autori propongono di addestrare più agenti distinti, ciascuno su un diverso sottoinsieme del training set.", "Gli autori propongono un approccio ensemble per la riformulazione delle query"]} +{"source": "Network Embeddings (NEs) map the nodes of a given network into $d$-dimensional Euclidean space $\\mathbb{R}^d$. Ideally, this mapping is such that 'similar' nodes are mapped onto nearby points, such that the NE can be used for purposes such as link prediction (if 'similar' means being 'more likely to be connected') or classification (if 'similar' means 'being more likely to have the same label'). In recent years various methods for NE have been introduced, all following a similar strategy: defining a notion of similarity between nodes (typically some distance measure within the network), a distance measure in the embedding space, and a loss function that penalizes large distances for similar nodes and small distances for dissimilar nodes. A difficulty faced by existing methods is that certain networks are fundamentally hard to embed due to their structural properties: (approximate) multipartiteness, certain degree distributions, assortativity, etc. To overcome this, we introduce a conceptual innovation to the NE literature and propose to create \\emph{Conditional Network Embeddings} (CNEs); embeddings that maximally add information with respect to given structural properties (e.g. node degrees, block densities, etc.). We use a simple Bayesian approach to achieve this, and propose a block stochastic gradient descent algorithm for fitting it efficiently. We demonstrate that CNEs are superior for link prediction and multi-label classification when compared to state-of-the-art methods, and this without adding significant mathematical or computational complexity. Finally, we illustrate the potential of CNE for network visualization.", "target": ["Introduciamo un metodo di embedding della rete che tiene conto delle informazioni del prior sulla rete, ottenendo prestazioni empiriche superiori.", "L'articolo ha proposto di utilizzare una distribuzione a priori per vincolare l'embedding della rete, per la formulazione questo articolo ha utilizzato distribuzioni gaussiane molto ristrette.", "Propone l'apprendimento unsupervised degli embedding dei nodi considerando le proprietà strutturali delle reti."]} +{"source": "This paper studies a class of adaptive gradient based momentum algorithms that update the search directions and learning rates simultaneously using past gradients. This class, which we refer to as the ''``Adam-type'', includes the popular algorithms such as Adam, AMSGrad, AdaGrad. Despite their popularity in training deep neural networks (DNNs), the convergence of these algorithms for solving non-convex problems remains an open question. In this paper, we develop an analysis framework and a set of mild sufficient conditions that guarantee the convergence of the Adam-type methods, with a convergence rate of order $O(\\log{T}/\\sqrt{T})$ for non-convex stochastic optimization. Our convergence analysis applies to a new algorithm called AdaFom (AdaGrad with First Order Momentum). We show that the conditions are essential, by identifying concrete examples in which violating the conditions makes an algorithm diverge. Besides providing one of the first comprehensive analysis for Adam-type methods in the non-convex setting, our results can also help the practitioners to easily monitor the progress of algorithms and determine their convergence behavior.", "target": ["Analizziamo la convergenza degli algoritmi di tipo Adam e forniamo lievi condizioni sufficienti per garantire la loro convergenza, mostriamo anche che la violazione delle condizioni può far divergere un algoritmo.", "Presenta un'analisi della convergenza in un ambiente non convesso per una famiglia di algoritmi di ottimizzazione.", "Questo articolo studia la condizione di convergenza degli ottimizzatori di tipo Adam nei problemi di ottimizzazione non vincolati non convessi."]} +{"source": "This research paper describes a simplistic architecture named as AANN: Absolute Artificial Neural Network, which can be used to create highly interpretable representations of the input data. These representations are generated by penalizing the learning of the network in such a way that those learned representations correspond to the respective labels present in the labelled dataset used for supervised training; thereby, simultaneously giving the network the ability to classify the input data. The network can be used in the reverse direction to generate data that closely resembles the input by feeding in representation vectors as required. This research paper also explores the use of mathematical abs (absolute valued) functions as activation functions which constitutes the core part of this neural network architecture. Finally the results obtained on the MNIST dataset by using this technique are presented and discussed in brief.", "target": ["L'auto-encoder a pesi vincolati con la funzione di valore assoluto come funzione di attivazione, impara a fare la classificazione nella forward direction e la regressione nella backward direction grazie alla funzione di costo appositamente definita.", "L'articolo propone di utilizzare la funzione di attivazione del valore assoluto in un'architettura autoencoder con un termine aggiuntivo di apprendimento supervisionato nella funzione obiettivo", "Questo paper introduce una rete reversibile con il valore assoluto usato come funzione di attivazione"]} +{"source": "Current state-of-the-art relation extraction methods typically rely on a set of lexical, syntactic, and semantic features, explicitly computed in a pre-processing step. Training feature extraction models requires additional annotated language resources, which severely restricts the applicability and portability of relation extraction to novel languages. Similarly, pre-processing introduces an additional source of error. To address these limitations, we introduce TRE, a Transformer for Relation Extraction, extending the OpenAI Generative Pre-trained Transformer [Radford et al., 2018]. Unlike previous relation extraction models, TRE uses pre-trained deep language representations instead of explicit linguistic features to inform the relation classification and combines it with the self-attentive Transformer architecture to effectively model long-range dependencies between entity mentions. TRE allows us to learn implicit linguistic features solely from plain text corpora by unsupervised pre-training, before fine-tuning the learned language representations on the relation extraction task. TRE obtains a new state-of-the-art result on the TACRED and SemEval 2010 Task 8 datasets, achieving a test F1 of 67.4 and 87.1, respectively. Furthermore, we observe a significant increase in sample efficiency. With only 20% of the training examples, TRE matches the performance of our baselines and our model trained from scratch on 100% of the TACRED dataset. We open-source our trained models, experiments, and source code.", "target": ["Proponiamo un modello di relation extraction basato su Transformer che usa rappresentazioni linguistiche pre-trained invece di feature linguistiche esplicite.", "Presenta un modello di relation extraction basato su trasformer che sfrutta il pre-training su testo non annotato con un obiettivo di language modeling.", "Questo articolo descrive una nuova applicazione di reti di trasformer per relation extraction.", "L'articolo presenta un'architettura basata su Transformer per relation extraction, valutandola su due dataset."]} +{"source": "Neural networks have recently had a lot of success for many tasks. However, neural network architectures that perform well are still typically designed manually by experts in a cumbersome trial-and-error process. We propose a new method to automatically search for well-performing CNN architectures based on a simple hill climbing procedure whose operators apply network morphisms, followed by short optimization runs by cosine annealing. Surprisingly, this simple method yields competitive results, despite only requiring resources in the same order of magnitude as training a single network. E.g., on CIFAR-10, our method designs and trains networks with an error rate below 6% in only 12 hours on a single GPU; training for one day reduces this error further, to almost 5%.", "target": ["Proponiamo un metodo semplice ed efficace per la ricerca di architetture per le reti neurali convoluzionali.", "Propone un metodo di ricerca dell'architettura neurale che raggiunge una precisione vicina allo stato dell'arte su CIFAR10 e richiede molte meno risorse computazionali.", "Presenta un metodo per cercare architetture di reti neurali contemporanemente al training che risparmia drasticamente sul tempo di training e sul tempo di ricerca dell'architettura.", "Propone una variante di ricerca dell'architettura neurale usando morfismi di rete per definire uno spazio di ricerca usando architetture CNN che completano il task di classificazione delle immagini CIFAR"]} +{"source": "We propose GraphGAN - the first implicit generative model for graphs that enables to mimic real-world networks. We pose the problem of graph generation as learning the distribution of biased random walks over a single input graph. Our model is based on a stochastic neural network that generates discrete output samples, and is trained using the Wasserstein GAN objective. GraphGAN enables us to generate sibling graphs, which have similar properties yet are not exact replicas of the original graph. Moreover, GraphGAN learns a semantic mapping from the latent input space to the generated graph's properties. We discover that sampling from certain regions of the latent space leads to varying properties of the output graphs, with smooth transitions between them. Strong generalization properties of GraphGAN are highlighted by its competitive performance in link prediction as well as promising results on node classification, even though not specifically trained for these tasks.", "target": ["Si utilizzano le GAN per generare grafi tramite random walk.", "Gli autori hanno proposto un modello generativo di random walk su grafi che permette l'apprendimento model-agnostic, il fitting controllabile, la generazione di ensemble di grafi", "Propone una formulazione WGAN per generare grafi basati su random walk usando gli embedding dei nodi e un'architettura LSTM per la modellazione."]} +{"source": "The ability of a classifier to recognize unknown inputs is important for many classification-based systems. We discuss the problem of simultaneous classification and novelty detection, i.e. determining whether an input is from the known set of classes and from which specific class, or from an unknown domain and does not belong to any of the known classes. We propose a method based on the Generative Adversarial Networks (GAN) framework. We show that a multi-class discriminator trained with a generator that generates samples from a mixture of nominal and novel data distributions is the optimal novelty detector. We approximate that generator with a mixture generator trained with the Feature Matching loss and empirically show that the proposed method outperforms conventional methods for novelty detection. Our findings demonstrate a simple, yet powerful new application of the GAN framework for the task of novelty detection.", "target": ["Proponiamo di risolvere un problema di classificazione simultanea e novelty detection nel framework GAN.", "Propone una GAN per unificare la classificazione e la novelty detection.", "L'articolo presenta un metodo per la novelty detection basato su una GAN multiclasse che è addestrata a produrre immagini generate da una mistura di distribuzioni nominali e nuove.", "L'articolo propone una GAN per la novelty detection utilizzando un generatore di misture con loss di feature matching"]} +{"source": "Verifying a person's identity based on their voice is a challenging, real-world problem in biometric security. A crucial requirement of such speaker verification systems is to be domain robust. Performance should not degrade even if speakers are talking in languages not seen during training. To this end, we present a flexible and interpretable framework for learning domain invariant speaker embeddings using Generative Adversarial Networks. We combine adversarial training with an angular margin loss function, which encourages the speaker embedding model to be discriminative by directly optimizing for cosine similarity between classes. We are able to beat a strong baseline system using a cosine distance classifier and a simple score-averaging strategy. Our results also show that models with adversarial adaptation perform significantly better than unadapted models. In an attempt to better understand this behavior, we quantitatively measure the degree of invariance induced by our proposed methods using Maximum Mean Discrepancy and Frechet distances. Our analysis shows that our proposed adversarial speaker embedding models significantly reduce the distance between source and target data distributions, while performing similarly on the former and better on the latter.", "target": ["Le prestazioni di speaker verification possono essere migliorate significativamente adattando il modello ai dati in-domain utilizzando le generative adversarial network. Inoltre, l'adattamento può essere eseguito in modo unsupervised.", "Propone una serie di varianti di GAN per il task di speaker verification nella condizione di dominio non corrispondente."]} +{"source": "Learning disentangling representations of the independent factors of variations that explain the data in an unsupervised setting is still a major challenge. In the following paper we address the task of disentanglement and introduce a new state-of-the-art approach called Non-synergistic variational Autoencoder (Non-Syn VAE). Our model draws inspiration from population coding, where the notion of synergy arises when we describe the encoded information by neurons in the form of responses from the stimuli. If those responses convey more information together than separate as independent sources of encoding information, they are acting synergetically. By penalizing the synergistic mutual information within the latents we encourage information independence and by doing that disentangle the latent factors. Notably, our approach could be added to the VAE framework easily, where the new ELBO function is still a lower bound on the log likelihood. In addition, we qualitatively compare our model with Factor VAE and show that this one implicitly minimises the synergy of the latents.", "target": ["Minimizzare l'informazione mutua sinergica tra i latenti e i dati per il task di disentanglement usando il framework VAE.", "Propone una nuova funzione obiettivo per l'apprendimento di rappresentazioni disentangled in un framework variazionale minimizzando la sinergia delle informazioni fornite.", "Gli autori mirano ad addestrare un VAE che ha dissociato le rappresentazioni latenti in modo \"sinergicamente\" massimo. ", "Questo articolo propone un nuovo approccio per imporre il disentanglement nei VAE utilizzando un termine che penalizza l'informazione mutua sinergica tra le variabili latenti."]} +{"source": "Metric embeddings are immensely useful representations of associations between entities (images, users, search queries, words, and more). Embeddings are learned by optimizing a loss objective of the general form of a sum over example associations. Typically, the optimization uses stochastic gradient updates over minibatches of examples that are arranged independently at random. In this work, we propose the use of {\\em structured arrangements} through randomized {\\em microbatches} of examples that are more likely to include similar ones. We make a principled argument for the properties of our arrangements that accelerate the training and present efficient algorithms to generate microbatches that respect the marginal distribution of training examples. Finally, we observe experimentally that our structured arrangements accelerate training by 3-20\\%. Structured arrangements emerge as a powerful and novel performance knob for SGD that is independent and complementary to other SGD hyperparameters and thus is a candidate for wide deployment.", "target": ["Accelerare SGD disponendo gli esempi in modo diverso", "L'articolo presenta un metodo per migliorare il tasso di convergenza di Stochastic Gradient Descent per l'apprendimento di embedding raggruppando sample di training simili.", "Propone una strategia di sampling non uniforme per costruire minibatch in SGD per il task di imparare embedding per associazioni di oggetti."]} +{"source": "Ubuntu dialogue corpus is the largest public available dialogue corpus to make it feasible to build end-to-end deep neural network models directly from the conversation data. One challenge of Ubuntu dialogue corpus is the large number of out-of-vocabulary words. In this paper we proposed an algorithm which combines the general pre-trained word embedding vectors with those generated on the task-specific training set to address this issue. We integrated character embedding into Chen et al's Enhanced LSTM method (ESIM) and used it to evaluate the effectiveness of our proposed method. For the task of next utterance selection, the proposed method has demonstrated a significant performance improvement against original ESIM and the new model has achieved state-of-the-art results on both Ubuntu dialogue corpus and Douban conversation corpus. In addition, we investigated the performance impact of end-of-utterance and end-of-turn token tags.", "target": ["Combina le informazioni tra i word embedding pre-costruiti e i word embedding specifici per il task per affrontare il problema dell'out-of-vocabulary ", "Questo articolo propone un approccio per migliorare la predizione di embedding out-of-vocabulary per il task di modeling dei dialoghi con guadagni considerevoli rispetto alle baseline.", "Propone di combinare gli embedding di parole esterne preaddestrati e gli embedding di parole preaddestrati sui dati di training mantenendoli come due view separate.", "Propone un metodo per estendere la copertura di word embedding pre-addestrati per affrontare il problema OOV che si presenta quando li si applica a dataset di conversazione e applica inoltre nuove varianti del modello basato su LSTM al task di response-selection nel dialogue modeling."]} +{"source": "Deep learning has shown that learned functions can dramatically outperform hand-designed functions on perceptual tasks. Analogously, this suggests that learned update functions may similarly outperform current hand-designed optimizers, especially for specific tasks. However, learned optimizers are notoriously difficult to train and have yet to demonstrate wall-clock speedups over hand-designed optimizers, and thus are rarely used in practice. Typically, learned optimizers are trained by truncated backpropagation through an unrolled optimization process. The resulting gradients are either strongly biased (for short truncations) or have exploding norm (for long truncations). In this work we propose a training scheme which overcomes both of these difficulties, by dynamically weighting two unbiased gradient estimators for a variational loss on optimizer performance. This allows us to train neural networks to perform optimization faster than well tuned first-order methods. Moreover, by training the optimizer against validation loss, as opposed to training loss, we are able to use it to train models which generalize better than those trained by first order methods. We demonstrate these results on problems where our learned optimizer trains convolutional networks in a fifth of the wall-clock time compared to tuned first-order methods, and with an improvement", "target": ["Analizziamo i problemi durante il training degli ottimizzatori appresi, affrontiamo questi problemi tramite l'ottimizzazione variazionale utilizzando due stimatori di gradiente complementari, e addestriamo ottimizzatori che sono 5 volte più veloci in termini di wall-clock time rispetto agli ottimizzatori di base (per esempio Adam).", "Questo articolo usa l'ottimizzazione un-rolled per imparare le reti neurali per l'ottimizzazione.", "Questo articolo affronta il problema dell'apprendimento di un ottimizzatore, in particolare gli autori si concentrano sull'ottenimento di gradienti più puliti dalla procedura di training unrolled.", "Presenta un metodo per \"imparare un ottimizzatore\" usando un'ottimizzazione variazionale per la loss \"esterna\" dell'ottimizzatore e propone l'idea di combinare sia il gradiente riparametrizzato che lo stimatore della funzione di scoring per l'obiettivo variazionale e li pondera usando un prodotto di Gaussiane per la media."]} +{"source": "Asynchronous distributed gradient descent algorithms for training of deep neural networks are usually considered as inefficient, mainly because of the Gradient delay problem. In this paper, we propose a novel asynchronous distributed algorithm that tackles this limitation by well-thought-out averaging of model updates, computed by workers. The algorithm allows computing gradients along the process of gradient merge, thus, reducing or even completely eliminating worker idle time due to communication overhead, which is a pitfall of existing asynchronous methods. We provide theoretical analysis of the proposed asynchronous algorithm, and show its regret bounds. According to our analysis, the crucial parameter for keeping high convergence rate is the maximal discrepancy between local parameter vectors of any pair of workers. As long as it is kept relatively small, the convergence rate of the algorithm is shown to be the same as the one of a sequential online learning. Furthermore, in our algorithm, this discrepancy is bounded by an expression that involves the staleness parameter of the algorithm, and is independent on the number of workers. This is the main differentiator between our approach and other solutions, such as Elastic Asynchronous SGD or Downpour SGD, in which that maximal discrepancy is bounded by an expression that depends on the number of workers, due to gradient delay problem. To demonstrate effectiveness of our approach, we conduct a series of experiments on image classification task on a cluster with 4 machines, equipped with a commodity communication switch and with a single GPU card per machine. Our experiments show a linear scaling on 4-machine cluster without sacrificing the test accuracy, while eliminating almost completely worker idle time. Since our method allows using commodity communication switch, it paves a way for large scale distributed training performed on commodity clusters.", "target": ["Un metodo per un efficiente training distribuito asincrono di modelli di deep learning insieme ai bound teorici di regret.", "L'articolo propone un algoritmo per limitare la staleness in SGD asincrono e fornisce un'analisi teorica", "Propone un algoritmo ibrido per eliminare il ritardo del gradiente nei metodi asincroni."]} +{"source": "Neural network quantization is becoming an industry standard to efficiently deploy deep learning models on hardware platforms, such as CPU, GPU, TPU, and FPGAs. However, we observe that the conventional quantization approaches are vulnerable to adversarial attacks. This paper aims to raise people's awareness about the security of the quantized models, and we designed a novel quantization methodology to jointly optimize the efficiency and robustness of deep learning models. We first conduct an empirical study to show that vanilla quantization suffers more from adversarial attacks. We observe that the inferior robustness comes from the error amplification effect, where the quantization operation further enlarges the distance caused by amplified noise. Then we propose a novel Defensive Quantization (DQ) method by controlling the Lipschitz constant of the network during quantization, such that the magnitude of the adversarial noise remains non-expansive during inference. Extensive experiments on CIFAR-10 and SVHN datasets demonstrate that our new quantization method can defend neural networks against adversarial examples, and even achieves superior robustness than their full-precision counterparts, while maintaining the same hardware efficiency as vanilla quantization approaches. As a by-product, DQ can also improve the accuracy of quantized models without adversarial attack.", "target": ["Abbiamo progettato una nuova metodologia di quantizzazione per ottimizzare congiuntamente l'efficienza e la robustezza dei modelli di deep learning.", "Propone uno schema di regolarizzazione per proteggere le reti neurali quantizzate dagli adversarial attack usando un filtraggio costante di Lipschitz dell'input-output dei layer interni."]} +{"source": "Recurrent Neural Networks (RNNs) continue to show outstanding performance in sequence modeling tasks. However, training RNNs on long sequences often face challenges like slow inference, vanishing gradients and difficulty in capturing long term dependencies. In backpropagation through time settings, these issues are tightly coupled with the large, sequential computational graph resulting from unfolding the RNN in time. We introduce the Skip RNN model which extends existing RNN models by learning to skip state updates and shortens the effective size of the computational graph. This model can also be encouraged to perform fewer state updates through a budget constraint. We evaluate the proposed model on various tasks and show how it can reduce the number of required RNN updates while preserving, and sometimes even improving, the performance of the baseline RNN models. Source code is publicly available at https://imatge-upc.github.io/skiprnn-2017-telecombcn/.", "target": ["Una modifica per le architetture RNN esistenti che permette loro di saltare gli aggiornamenti di stato mantenendo le prestazioni delle architetture originali.", "Propone il modello Skip RNN che permette a una rete ricorrente di saltare selettivamente l'aggiornamento del suo hidden state per alcuni input, portando a una riduzione dei calcoli in fase di test.", "Propone un nuovo modello RNN in cui sia l'input che l'aggiornamento dello stato delle celle ricorrenti sono saltati in modo adattivo in alcuni time step."]} +{"source": "We propose a fast second-order method that can be used as a drop-in replacement for current deep learning solvers. Compared to stochastic gradient descent (SGD), it only requires two additional forward-mode automatic differentiation operations per iteration, which has a computational cost comparable to two standard forward passes and is easy to implement. Our method addresses long-standing issues with current second-order solvers, which invert an approximate Hessian matrix every iteration exactly or by conjugate-gradient methods, procedures that are much slower than a SGD step. Instead, we propose to keep a single estimate of the gradient projected by the inverse Hessian matrix, and update it once per iteration with just two passes over the network. This estimate has the same size and is similar to the momentum variable that is commonly used in SGD. No estimate of the Hessian is maintained. We first validate our method, called CurveBall, on small problems with known solutions (noisy Rosenbrock function and degenerate 2-layer linear networks), where current deep learning solvers struggle. We then train several large models on CIFAR and ImageNet, including ResNet and VGG-f networks, where we demonstrate faster convergence with no hyperparameter tuning. We also show our optimiser's generality by testing on a large set of randomly-generated architectures.", "target": ["Un veloce solutore del secondo ordine per il deep learning che funziona su problemi di scala ImageNet senza tuning degli iperparametri", "Scegliere la direzione usando un singolo passo di gradient descent \"verso il passo di Newton\" da una stima originale, e poi prendere questa direzione invece del gradiente originale", "Un nuovo metodo di ottimizzazione approssimata del secondo ordine a basso costo computazionale che sostituisce il calcolo della matrice Hessiana con un singolo gradient step e una strategia di warm start."]} +{"source": "The recently presented idea to learn heuristics for combinatorial optimization problems is promising as it can save costly development. However, to push this idea towards practical implementation, we need better models and better ways of training. We contribute in both directions: we propose a model based on attention layers with benefits over the Pointer Network and we show how to train this model using REINFORCE with a simple baseline based on a deterministic greedy rollout, which we find is more efficient than using a value function. We significantly improve over recent learned heuristics for the Travelling Salesman Problem (TSP), getting close to optimal results for problems up to 100 nodes. With the same hyperparameters, we learn strong heuristics for two variants of the Vehicle Routing Problem (VRP), the Orienteering Problem (OP) and (a stochastic variant of) the Prize Collecting TSP (PCTSP), outperforming a wide range of baselines and getting results close to highly optimized and specialized algorithms.", "target": ["Modello basato sull'attention addestrato con REINFORCE con la greedy rollout baseline per imparare euristiche con risultati competitivi su TSP e altri problemi di routing", "Presenta un approccio basato sull'attention per imparare una policy per risolvere TSP e altri problemi di ottimizzazione combinatoria di routing.", "Questo articolo cerca di imparare l'euristica per risolvere problemi di ottimizzazione combinatoria"]} +{"source": "We propose an efficient online hyperparameter optimization method which uses a joint dynamical system to evaluate the gradient with respect to the hyperparameters. While similar methods are usually limited to hyperparameters with a smooth impact on the model, we show how to apply it to the probability of dropout in neural networks. Finally, we show its effectiveness on two distinct tasks.", "target": ["Un algoritmo per ottimizzare gli iper-parametri di regolarizzazione durante il training", "L'articolo propone un modo per reinizializzare y ad ogni aggiornamento di lambda e una procedura di clipping di y per mantenere la stabilità del sistema dinamico.", "Propone un algoritmo per l'ottimizzazione degli iperparametri che può essere visto come un'estensione di Franceschi 2017 in cui alcune stime vengono riavviate a caldo per aumentare la stabilità del metodo.", "Propone un'estensione di un metodo esistente per ottimizzare gli iperparametri di regolarizzazione."]} +{"source": "Ongoing innovations in recurrent neural network architectures have provided a steady influx of apparently state-of-the-art results on language modelling benchmarks. However, these have been evaluated using differing codebases and limited computational resources, which represent uncontrolled sources of experimental variation. We reevaluate several popular architectures and regularisation methods with large-scale automatic black-box hyperparameter tuning and arrive at the somewhat surprising conclusion that standard LSTM architectures, when properly regularised, outperform more recent models. We establish a new state of the art on the Penn Treebank and Wikitext-2 corpora, as well as strong baselines on the Hutter Prize dataset.", "target": ["Mostrare che le LSTM sono altrettanto buone o migliori delle recenti innovazioni per LM e che la valutazione dei modelli è spesso inaffidabile.", "Questo articolo descrive una convalida completa dei language model a livello di parola e carattere basati su LSTM, portando a un risultato significativo nel language modeling e ad una pietra miliare nel deep learning."]} +{"source": "Residual and skip connections play an important role in many current generative models. Although their theoretical and numerical advantages are understood, their role in speech enhancement systems has not been investigated so far. When performing spectral speech enhancement, residual connections are very similar in nature to spectral subtraction, which is the one of the most commonly employed speech enhancement approaches. Highway networks, on the other hand, can be seen as a combination of spectral masking and spectral subtraction. However, when using deep neural networks, such operations would normally happen in a transformed spectral domain, as opposed to traditional speech enhancement where all operations are often done directly on the spectrum. In this paper, we aim to investigate the role of residual and highway connections in deep neural networks for speech enhancement, and verify whether or not they operate similarly to their traditional, digital signal processing counterparts. We visualize the outputs of such connections, projected back to the spectral domain, in models trained for speech denoising, and show that while skip connections do not necessarily improve performance with regards to the number of parameters, they make speech enhancement models more interpretable.", "target": ["Mostriamo come l'uso delle skip connection può rendere i modelli di speech enhancement più interpretabili, poiché li fa usare meccanismi simili a quelli che sono stati esplorati nella letteratura DSP.", "Gli autori propongono di incorporare i blocchi Residual, Highway e Masking all'interno di una pipeline completamente convoluzionale per capire come viene eseguita l'inferenza iterativa dell'output e il mascheramento in un task di speech enhancement", "Gli autori interpretano le connessioni highway, residual e di masking.", "Gli autori generano il loro discorso rumoroso aggiungendo artificialmente del rumore da un dataset sul rumore ben noto ad un dataset sul discorso senza rumore meno conosciuto."]} +{"source": "Bayesian neural networks (BNNs) hold great promise as a flexible and principled solution to deal with uncertainty when learning from finite data. Among approaches to realize probabilistic inference in deep neural networks, variational Bayes (VB) is theoretically grounded, generally applicable, and computationally efficient. With wide recognition of potential advantages, why is it that variational Bayes has seen very limited practical use for BNNs in real applications? We argue that variational inference in neural networks is fragile: successful implementations require careful initialization and tuning of prior variances, as well as controlling the variance of Monte Carlo gradient estimates. We provide two innovations that aim to turn VB into a robust inference tool for Bayesian neural networks: first, we introduce a novel deterministic method to approximate moments in neural networks, eliminating gradient variance; second, we introduce a hierarchical prior for parameters and a novel Empirical Bayes procedure for automatically selecting prior variances. Combining these two innovations, the resulting method is highly efficient and robust. On the application of heteroscedastic regression we demonstrate good predictive performance over alternative approaches.", "target": ["Un metodo per eliminare la varianza del gradiente e fare automaticamente tuning dei prior per un training efficace delle reti neurali bayesiane", "Propone un nuovo approccio per eseguire l'inferenza variazionale deterministica per BNN feed-forward con specifiche funzioni di attivazione non lineari approssimando i momenti layerwise.", "L'articolo considera un approccio puramente deterministico all'apprendimento di approssimazioni posteriori variazionali per le reti neurali bayesiane."]} +{"source": "Skills learned through (deep) reinforcement learning often generalizes poorly across tasks and re-training is necessary when presented with a new task. We present a framework that combines techniques in formal methods with reinforcement learning (RL) that allows for the convenient specification of complex temporal dependent tasks with logical expressions and construction of new skills from existing ones with no additional exploration. We provide theoretical results for our composition technique and evaluate on a simple grid world simulation as well as a robotic manipulation task.", "target": ["Un metodo formale per la composizione di skill in task di reinforcement learning", "L'articolo combina il RL e vincoli espressi da formule logiche impostando un'automazione da formule scTLTL.", "Propone un metodo che aiuta a costruire la policy dai sottotask appresi sul tema della combinazione di task RL con formule logiche temporali lineari."]} +{"source": "The application of multi-modal generative models by means of a Variational Auto Encoder (VAE) is an upcoming research topic for sensor fusion and bi-directional modality exchange. This contribution gives insights into the learned joint latent representation and shows that expressiveness and coherence are decisive properties for multi-modal datasets. Furthermore, we propose a multi-modal VAE derived from the full joint marginal log-likelihood that is able to learn the most meaningful representation for ambiguous observations. Since the properties of multi-modal sensor setups are essential for our approach but hardly available, we also propose a technique to generate correlated datasets from uni-modal ones.", "target": ["Derivare una formulazione generale di un VAE multimodale dalla log-likelihood marginale congiunta.", "Propone un VAE multimodale con un bound variazionale derivato dalla regola della catena.", "Questo articolo propone un obiettivo, M^2VAE, per i VAE multimodali, che dovrebbe imparare una rappresentazione dello spazio latente più significativa."]} +{"source": "We build on auto-encoding sequential Monte Carlo (AESMC): a method for model and proposal learning based on maximizing the lower bound to the log marginal likelihood in a broad family of structured probabilistic models. Our approach relies on the efficiency of sequential Monte Carlo (SMC) for performing inference in structured probabilistic models and the flexibility of deep neural networks to model complex conditional probability distributions. We develop additional theoretical insights and introduce a new training procedure which improves both model and proposal learning. We demonstrate that our approach provides a fast, easy-to-implement and scalable means for simultaneous model learning and proposal adaptation in deep generative models.", "target": ["Ci basiamo sull'autoencoding sequenziale Monte Carlo, otteniamo nuove intuizioni teoriche e sviluppiamo una procedura di training migliorata basata su queste intuizioni.", "L'articolo propone una versione del training in stile IWAE che usa SMC invece del classico importance sampling.", "Questo lavoro propone l'autoencoding sequenziale Monte Carlo, estendendo il framework VAE a un nuovo obiettivo Monte Carto basato su SMC."]} +{"source": "A key component for many reinforcement learning agents is to learn a value function, either for policy evaluation or control. Many of the algorithms for learning values, however, are designed for linear function approximation---with a fixed basis or fixed representation. Though there have been a few sound extensions to nonlinear function approximation, such as nonlinear gradient temporal difference learning, these methods have largely not been adopted, eschewed in favour of simpler but not sound methods like temporal difference learning and Q-learning. In this work, we provide a two-timescale network (TTN) architecture that enables linear methods to be used to learn values, with a nonlinear representation learned at a slower timescale. The approach facilitates the use of algorithms developed for the linear setting, such as data-efficient least-squares methods, eligibility traces and the myriad of recently developed linear policy evaluation algorithms, to provide nonlinear value estimates. We prove convergence for TTNs, with particular care given to ensure convergence of the fast linear component under potentially dependent features provided by the learned representation. We empirically demonstrate the benefits of TTNs, compared to other nonlinear value function approximation algorithms, both for policy evaluation and control.", "target": ["Proponiamo un'architettura per l'apprendimento delle funzioni di valore che permette l'uso di qualsiasi algoritmo lineare di valutazione della policy in tandem con l'apprendimento non lineare delle feature.", "L'articolo propone un framework a due scale temporali per l'apprendimento della funzione di valore e della rappresentazione dello stato con approssimazioni non lineari.", "Questo articolo propone le reti a due scale temporali e prova la convergenza di questo metodo usando metodi di approssimazione stocastica a due scale temporali.", "Questo articolo presenta una rete a due scale temporali che permette di utilizzare metodi lineari per imparare i valori."]} +{"source": "Large-scale Long Short-Term Memory (LSTM) cells are often the building blocks of many state-of-the-art algorithms for tasks in Natural Language Processing (NLP). However, LSTMs are known to be computationally inefficient because the memory capacity of the models depends on the number of parameters, and the inherent recurrence that models the temporal dependency is not parallelizable. In this paper, we propose simple, but effective, low-rank matrix factorization (MF) algorithms to compress network parameters and significantly speed up LSTMs with almost no loss of performance (and sometimes even gain). To show the effectiveness of our method across different tasks, we examine two settings: 1) compressing core LSTM layers in Language Models, 2) compressing biLSTM layers of ELMo~\\citep{ELMo} and evaluate in three downstream NLP tasks (Sentiment Analysis, Textual Entailment, and Question Answering). The latter is particularly interesting as embeddings from large pre-trained biLSTM Language Models are often used as contextual word representations. Finally, we discover that matrix factorization performs better in general, additive recurrence is often more important than multiplicative recurrence, and we identify an interesting correlation between matrix norms and compression performance.", "target": ["Proponiamo semplici, ma efficaci, algoritmi di fattorizzazione di matrice a basso rango (MF) per accelerare il tempo di esecuzione, risparmiare memoria e migliorare le prestazioni delle LSTM.", "Propone di accelerare le LSTM usando MF come strategia di compressione post-elaborazione e conduce ampi esperimenti per mostrare le prestazioni."]} +{"source": "Manipulation and re-use of images in scientific publications is a recurring problem, at present lacking a scalable solution. Existing tools for detecting image duplication are mostly manual or semi-automated, despite the fact that generating data for a learning-based approach is straightforward, as we here illustrate. This paper addresses the problem of determining if, given two images, one is a manipulated version of the other by means of certain geometric and statistical manipulations, e.g. copy, rotation, translation, scale, perspective transform, histogram adjustment, partial erasing, and compression artifacts. We propose a solution based on a 3-branch Siamese Convolutional Neural Network. The ConvNet model is trained to map images into a 128-dimensional space, where the Euclidean distance between duplicate (respectively, unique) images is no greater (respectively, greater) than 1. Our results suggest that such an approach can serve as tool to improve surveillance of the published and in-peer-review literature for image manipulation. We also show that as a byproduct the network learns useful representations for semantic segmentation, with performance comparable to that of domain-specific models.", "target": ["Una metrica forense per determinare se una data immagine è una copia (con possibile manipolazione) di un'altra immagine da un dato dataset.", "Introduce la siamese network per identificare le immagini duplicate e copiate/modificate, che può essere usata per migliorare la sorveglianza della letteratura pubblicata e in peer-review.", "L'articolo presenta un'applicazione di deep convolutional network per il task di rilevamento di immagini duplicate", "Questo lavoro affronta il problema di trovare immagini duplicate/near duplicate da pubblicazioni biomediche e propone una CNN standard e funzioni di loss, applicandole a questo campo."]} +{"source": "Training generative adversarial networks is unstable in high-dimensions as the true data distribution tends to be concentrated in a small fraction of the ambient space. The discriminator is then quickly able to classify nearly all generated samples as fake, leaving the generator without meaningful gradients and causing it to deteriorate after a point in training. In this work, we propose training a single generator simultaneously against an array of discriminators, each of which looks at a different random low-dimensional projection of the data. Individual discriminators, now provided with restricted views of the input, are unable to reject generated samples perfectly and continue to provide meaningful gradients to the generator throughout training. Meanwhile, the generator learns to produce samples consistent with the full data distribution to satisfy all discriminators simultaneously. We demonstrate the practical utility of this approach experimentally, and show that it is able to produce image samples with higher quality than traditional training with a single discriminator.", "target": ["Training stabile delle GAN ad alte dimensioni utilizzando un array di discriminatori, ciascuno con una visione a bassa dimensione dei sample generati", "L'articolo propone di stabilizzare il training delle GAN utilizzando un insieme di discriminatori, ognuno dei quali lavora su una proiezione casuale dei dati di input, per fornire il segnale di training per il modello generatore.", "L'articolo propone un metodo di training delle GAN per migliorare la stabilità del training.", "L'articolo propone un nuovo approccio al training delle GAN, che fornisce gradienti stabili per addestrare il generatore."]} +{"source": "We present a novel method to precisely impose tree-structured category information onto word-embeddings, resulting in ball embeddings in higher dimensional spaces (N-balls for short). Inclusion relations among N-balls implicitly encode subordinate relations among categories. The similarity measurement in terms of the cosine function is enriched by category information. Using a geometric construction method instead of back-propagation, we create large N-ball embeddings that satisfy two conditions: (1) category trees are precisely imposed onto word embeddings at zero energy cost; (2) pre-trained word embeddings are well preserved. A new benchmark data set is created for validating the category of unknown words. Experiments show that N-ball embeddings, carrying category information, significantly outperform word embeddings in the test of nearest neighborhoods, and demonstrate surprisingly good performance in validating categories of unknown words. Source codes and data-sets are free for public access \\url{https://github.com/gnodisnait/nball4tree.git} and \\url{https://github.com/gnodisnait/bp94nball.git}.", "target": ["Mostriamo un metodo geometrico per codificare perfettamente le informazioni dell'albero delle categorie in word-embedding pre-addestrati.", "L'articolo propone N-ball embedding per i dati tassonomici dove una N-ball è una coppia di un vettore centroide e il raggio dal centro.", "L'articolo presenta un metodo per modificare gli embedding vettoriali esistenti di oggetti categorici (come le parole), per convertirle in embedding a sfera che seguono le gerarchie.", "Si concentra sulla regolazione dei pretrained word embedding in modo che rispettino la relazione ipernimia/iponimia mediante un appropriato incapsulamento n-ball."]} +{"source": "For the challenging semantic image segmentation task the best performing models have traditionally combined the structured modelling capabilities of Conditional Random Fields (CRFs) with the feature extraction power of CNNs. In more recent works however, CRF post-processing has fallen out of favour. We argue that this is mainly due to the slow training and inference speeds of CRFs, as well as the difficulty of learning the internal CRF parameters. To overcome both issues we propose to add the assumption of conditional independence to the framework of fully-connected CRFs. This allows us to reformulate the inference in terms of convolutions, which can be implemented highly efficiently on GPUs.Doing so speeds up inference and training by two orders of magnitude. All parameters of the convolutional CRFs can easily be optimized using backpropagation. Towards the goal of facilitating further CRF research we have made our implementations publicly available.", "target": ["Proponiamo Convolutional CRF, un'alternativa veloce, potente e allenabile ai Fully Connected CRF.", "Gli autori sostituiscono il grande passo di filtraggio nel lattice permutoedrico con un kernel convoluzionario spazialmente variabile e mostrano che l'inferenza è più efficiente e il training più facile.", "Propone di eseguire il message passing su un CRF a kernel gaussiano troncato usando un kernel definito e il message passing parallelizzato su GPU."]} +{"source": "Deep Learning NLP domain lacks procedures for the analysis of model robustness. In this paper we propose a framework which validates robustness of any Question Answering model through model explainers. We propose that output of a robust model should be invariant to alterations that do not change its semantics. We test this property by manipulating question in two ways: swapping important question word for 1) its semantically correct synonym and 2) for word vector that is close in embedding space. We estimate importance of words in asked questions with Locally Interpretable Model Agnostic Explanations method (LIME). With these two steps we compare state-of-the-art Q&A models. We show that although accuracy of state-of-the-art models is high, they are very fragile to changes in the input. We can choose architecture that is more immune to attacks and thus more robust and stable in production environment. Morevoer, we propose 2 adversarial training scenarios which raise model sensitivity to true synonyms by up to 7% accuracy measure. Our findings help to understand which models are more stable and how they can be improved. In addition, we have created and published a new dataset that may be used for validation of robustness of a Q&A model.", "target": ["Proponiamo un approccio model agnostic per la convalida della robustezza del sistema Q&A e dimostriamo i risultati su modelli Q&A allo stato dell'arte.", "Affronta il problema della robustezza alle informazioni adversarial nel question answering.", "Migliorare la robustezza della comprensione automatica/question answering."]} +{"source": "In this paper, we propose a mix-generator generative adversarial networks (PGAN) model that works in parallel by mixing multiple disjoint generators to approximate a complex real distribution. In our model, we propose an adjustment component that collects all the generated data points from the generators, learns the boundary between each pair of generators, and provides error to separate the support of each of the generated distributions. To overcome the instability in a multiplayer game, a shrinkage adjustment component method is introduced to gradually reduce the boundary between generators during the training procedure. To address the linearly growing training time problem in a multiple generators model, we propose a method to train the generators in parallel. This means that our work can be scaled up to large parallel computation frameworks. We present an efficient loss function for the discriminator, an effective adjustment component, and a suitable generator. We also show how to introduce the decay factor to stabilize the training procedure. We have performed extensive experiments on synthetic datasets, MNIST, and CIFAR-10. These experiments reveal that the error provided by the adjustment component could successfully separate the generated distributions and each of the generators can stably learn a part of the real distribution even if only a few modes are contained in the real distribution.", "target": ["Multi generatore per catturare Pdata, risolvere la concorrenza e il problema di one-beat-all", "Propone GAN parallele per evitare il mode collapse nelle GAN attraverso una combinazione di più generatori deboli."]} +{"source": "We capitalize on the natural compositional structure of images in order to learn object segmentation with weakly labeled images. The intuition behind our approach is that removing objects from images will yield natural images, however removing random patches will yield unnatural images. We leverage this signal to develop a generative model that decomposes an image into layers, and when all layers are combined, it reconstructs the input image. However, when a layer is removed, the model learns to produce a different image that still looks natural to an adversary, which is possible by removing objects. Experiments and visualizations suggest that this model automatically learns object segmentation on images labeled only by scene better than baselines.", "target": ["Segmentazione di immagini debolmente supervisionata usando la struttura compositiva delle immagini e modelli generativi.", "Questo articolo crea una rappresentazione a layer per imparare meglio la segmentazione da immagini senza label.", "Questo articolo propone un modello generativo basato su GAN che decompone le immagini in layer multipli, dove l'obiettivo della GAN è quello di distinguere le immagini reali dalle immagini formate combinando i layer.", "Questo articolo propone un'architettura di rete neurale intorno all'idea della composizione stratificata della scena"]} +{"source": "Adversarial examples are a pervasive phenomenon of machine learning models where seemingly imperceptible perturbations to the input lead to misclassifications for otherwise statistically accurate models. We propose a geometric framework, drawing on tools from the manifold reconstruction literature, to analyze the high-dimensional geometry of adversarial examples. In particular, we highlight the importance of codimension: for low-dimensional data manifolds embedded in high-dimensional space there are many directions off the manifold in which to construct adversarial examples. Adversarial examples are a natural consequence of learning a decision boundary that classifies the low-dimensional data manifold well, but classifies points near the manifold incorrectly. Using our geometric framework we prove (1) a tradeoff between robustness under different norms, (2) that adversarial training in balls around the data is sample inefficient, and (3) sufficient sampling conditions under which nearest neighbor classifiers and ball-based adversarial training are robust.", "target": ["Presentiamo un framework geometrico per dimostrare le garanzie di robustezza ed evidenziare l'importanza della codimensione negli adversarial example.", "Questo articolo fornisce un'analisi teorica degli adversarial example, mostrando che esiste un compromesso tra la robustezza in diverse norme, l'adversarial training è sample inefficient, e il classificatore nearest neighbor può essere robusto in certe condizioni."]} +{"source": "Character-based neural machine translation (NMT) models alleviate out-of-vocabulary issues, learn morphology, and move us closer to completely end-to-end translation systems. Unfortunately, they are also very brittle and easily falter when presented with noisy data. In this paper, we confront NMT models with synthetic and natural sources of noise. We find that state-of-the-art models fail to translate even moderately noisy texts that humans have no trouble comprehending. We explore two approaches to increase model robustness: structure-invariant word representations and robust training on noisy texts. We find that a model based on a character convolutional neural network is able to simultaneously learn representations robust to multiple kinds of noise.", "target": ["CharNMT è fragile", "Questo articolo studia l'impatto del rumore a livello di carattere su 4 diversi sistemi di traduzione automatica neurale", "Questo articolo studia empiricamente le prestazioni dei sistemi NMT a livello di carattere di fronte al rumore a livello di carattere, sia sintetizzato che naturale.", "Questo articolo studia l'impatto dell'input rumoroso sulla traduzione automatica e prova modi per rendere i modelli NMT più robusti"]} +{"source": "As neural networks grow deeper and wider, learning networks with hard-threshold activations is becoming increasingly important, both for network quantization, which can drastically reduce time and energy requirements, and for creating large integrated systems of deep networks, which may have non-differentiable components and must avoid vanishing and exploding gradients for effective learning. However, since gradient descent is not applicable to hard-threshold functions, it is not clear how to learn them in a principled way. We address this problem by observing that setting targets for hard-threshold hidden units in order to minimize loss is a discrete optimization problem, and can be solved as such. The discrete optimization goal is to find a set of targets such that each unit, including the output, has a linearly separable problem to solve. Given these targets, the network decomposes into individual perceptrons, which can then be learned with standard convex approaches. Based on this, we develop a recursive mini-batch algorithm for learning deep hard-threshold networks that includes the popular but poorly justified straight-through estimator as a special case. Empirically, we show that our algorithm improves classification accuracy in a number of settings, including for AlexNet and ResNet-18 on ImageNet, when compared to the straight-through estimator.", "target": ["Impariamo deep network con unità hard-threshold impostando gli obiettivi delle hidden unit utilizzando l'ottimizzazione combinatoria e i pesi tramite l'ottimizzazione convessa, con conseguente miglioramento delle prestazioni su ImageNet.", "L'articolo spiega e generalizza gli approcci per l'apprendimento delle reti neurali con attivazione hard.", "Questo articolo esamina il problema dell'ottimizzazione delle deep network con unità hard-threshold.", "L'articolo discute il problema dell'ottimizzazione delle reti neurali con hard-threshold e propone una nuova soluzione con un insieme di euristiche/approssimazioni."]} +{"source": "The robust and efficient recognition of visual relations in images is a hallmark of biological vision. Here, we argue that, despite recent progress in visual recognition, modern machine vision algorithms are severely limited in their ability to learn visual relations. Through controlled experiments, we demonstrate that visual-relation problems strain convolutional neural networks (CNNs). The networks eventually break altogether when rote memorization becomes impossible such as when the intra-class variability exceeds their capacity. We further show that another type of feedforward network, called a relational network (RN), which was shown to successfully solve seemingly difficult visual question answering (VQA) problems on the CLEVR datasets, suffers similar limitations. Motivated by the comparable success of biological vision, we argue that feedback mechanisms including working memory and attention are the key computational components underlying abstract visual reasoning.", "target": ["Utilizzando una nuova challenge controllata di visual-relation, dimostriamo che i task same-different mettono a dura prova la capacità delle CNN; sosteniamo che le visual-relation possono essere risolte meglio utilizzando strategie attentivo-mnemoniche.", "Dimostra che le reti neurali convoluzionali e relazionali non riescono a risolvere i problemi di visual-relation addestrando le reti su dati di visual-relation generati artificialmente.", "Questo articolo esplora come le attuali CNN e le reti relazionali non riescano a riconoscere le visual-relation nelle immagini."]} +{"source": "Visual Active Tracking (VAT) aims at following a target object by autonomously controlling the motion system of a tracker given visual observations. Previous work has shown that the tracker can be trained in a simulator via reinforcement learning and deployed in real-world scenarios. However, during training, such a method requires manually specifying the moving path of the target object to be tracked, which cannot ensure the tracker’s generalization on the unseen object moving patterns. To learn a robust tracker for VAT, in this paper, we propose a novel adversarial RL method which adopts an Asymmetric Dueling mechanism, referred to as AD-VAT. In AD-VAT, both the tracker and the target are approximated by end-to-end neural networks, and are trained via RL in a dueling/competitive manner: i.e., the tracker intends to lockup the target, while the target tries to escape from the tracker. They are asymmetric in that the target is aware of the tracker, but not vice versa. Specifically, besides its own observation, the target is fed with the tracker’s observation and action, and learns to predict the tracker’s reward as an auxiliary task. We show that such an asymmetric dueling mechanism produces a stronger target, which in turn induces a more robust tracker. To stabilize the training, we also propose a novel partial zero-sum reward for the tracker/target. The experimental results, in both 2D and 3D environments, demonstrate that the proposed method leads to a faster convergence in training and yields more robust tracking behaviors in different testing scenarios. For supplementary videos, see: https://www.youtube.com/playlist?list=PL9rZj4Mea7wOZkdajK1TsprRg8iUf51BS The code is available at https://github.com/zfw1226/active_tracking_rl", "target": ["Proponiamo AD-VAT, dove il tracker e l'oggetto target, visti come due agenti apprendibili, sono avversari e possono migliorarsi reciprocamente durante l'allenamento.", "Questo lavoro mira ad affrontare il problema dell'active tracking visivo con un meccanismo di training in cui l'inseguitore e il bersaglio sono avversari reciproci", "Questo articolo presenta un semplice task di deep RL multi-agente in cui un inseguitore in movimento cerca di seguire un bersaglio in movimento.", "Propone una nuova funzione di reward \"a somma parziale zero\", che incoraggia solo la competizione inseguitore-obiettivo quando sono vicini e penalizza quando sono troppo lontani."]} +{"source": "Identifying the hypernym relations that hold between words is a fundamental task in NLP. Word embedding methods have recently shown some capability to encode hypernymy. However, such methods tend not to explicitly encode the hypernym hierarchy that exists between words. In this paper, we propose a method to learn a hierarchical word embedding in a specific order to capture the hypernymy. To learn the word embeddings, the proposed method considers not only the hypernym relations that exists between words on a taxonomy, but also their contextual information in a large text corpus. The experimental results on a supervised hypernymy detection and a newly-proposed hierarchical path completion tasks show the ability of the proposed method to encode the hierarchy. Moreover, the proposed method outperforms previously proposed methods for learning word and hypernym-specific word embeddings on multiple benchmarks.", "target": ["Abbiamo presentato un metodo per imparare congiuntamente un Hierarchical Word Embedding (HWE) usando un corpus e una tassonomia per identificare le relazioni di iperonimia tra le parole.", "L'articolo presenta un metodo per imparare congiuntamente word embedding usando statistiche di co-occorrenza e incorporando informazioni gerarchiche dalle reti semantiche.", "Questo articolo ha proposto un metodo di apprendimento congiunto degli iperonimi sia dal testo grezzo che dai dati della tassonomia supervisionata.", "Questo articolo propone di aggiungere una misura di differenza di \"inclusione distributiva\" all'obiettivo GloVE allo scopo di rappresentare le relazioni di iperonimia."]} +{"source": "While self-organizing principles have motivated much of early learning models, such principles have rarely been included in deep learning architectures. Indeed, from a supervised learning perspective it seems that topographic constraints are rather decremental to optimal performance. Here we study a network model that incorporates self-organizing maps into a supervised network and show how gradient learning results in a form of a self-organizing learning rule. Moreover, we show that such a model is robust in the sense of its application to a variety of areas, which is believed to be a hallmark of biological learning systems.", "target": ["Integrazione di auto-organizzazione e apprendimento supervisionato in una rete neurale gerarchica", "L'articolo discute l'apprendimento in una rete neurale a tre layer, dove il layer intermedio è organizzato topograficamente e studia l'interazione tra l'apprendimento unsupervised e quello gerarchico supervisionato nel contesto biologico.", "Una variante supervisionata della mappa auto-organizzante di Kohonen, ma dove il layer di output lineare è sostituito con errore quadratico da un layer softmax con cross-entropy.", "Propone un modello che utilizza hidden neuron con funzioni di attivazione auto-organizzanti, i cui output alimentano un classificatore con funzione di output softmax."]} +{"source": "Quantization of a neural network has an inherent problem called accumulated quantization error, which is the key obstacle towards ultra-low precision, e.g., 2- or 3-bit precision. To resolve this problem, we propose precision highway, which forms an end-to-end high-precision information flow while performing the ultra-low-precision computation. First, we describe how the precision highway reduce the accumulated quantization error in both convolutional and recurrent neural networks. We also provide the quantitative analysis of the benefit of precision highway and evaluate the overhead on the state-of-the-art hardware accelerator. In the experiments, our proposed method outperforms the best existing quantization methods while offering 3-bit weight/activation quantization with no accuracy loss and 2-bit quantization with a 2.45 % top-1 accuracy loss in ResNet-50. We also report that the proposed method significantly outperforms the existing method in the 2-bit quantization of an LSTM for language modeling.", "target": ["Precision highway: un concetto generalizzato di flusso di informazioni ad alta precisione per la quantizzazione sub 4-bit ", "Indaga il problema della quantizzazione delle reti neurali impiegando una precision highway end-to-end per ridurre l'errore di quantizzazione accumulato e consentire una precisione ultra-bassa nelle deep neural network.", "Questo articolo studia i metodi per migliorare le prestazioni delle reti neurali quantizzate", "Questo articolo propone di mantenere un alto flusso di attivazione/gradiente in due tipi di strutture di reti, ResNet e LSTM."]} +{"source": "The vast majority of natural sensory data is temporally redundant. For instance, video frames or audio samples which are sampled at nearby points in time tend to have similar values. Typically, deep learning algorithms take no advantage of this redundancy to reduce computations. This can be an obscene waste of energy. We present a variant on backpropagation for neural networks in which computation scales with the rate of change of the data - not the rate at which we process the data. We do this by implementing a form of Predictive Coding wherein neurons communicate a combination of their state, and their temporal change in state, and quantize this signal using Sigma-Delta modulation. Intriguingly, this simple communication rule give rise to units that resemble biologically-inspired leaky integrate-and-fire neurons, and to a spike-timing-dependent weight-update similar to Spike-Timing Dependent Plasticity (STDP), a synaptic learning rule observed in the brain. We demonstrate that on MNIST, on a temporal variant of MNIST, and on Youtube-BB, a dataset with videos in the wild, our algorithm performs about as well as a standard deep network trained with backpropagation, despite only communicating discrete values between layers.", "target": ["Un algoritmo per addestrare reti neurali in modo efficiente su dati temporalmente ridondanti.", "L'articolo descrive uno schema di codifica neurale per l'apprendimento basato sugli spike nelle deep neural network", "Questo articolo presenta un metodo per l'apprendimento basato sugli spike che mira a ridurre il calcolo necessario durante l'apprendimento e il test quando si classificano dati temporalmente ridondanti.", "Questo articolo applica una versione di codifica predittiva dello schema di encoding Sigma-Delta per ridurre il carico computazionale su una rete di deep learning, combinando i tre componenti in un modo non visto in precedenza."]} +{"source": "Information bottleneck (IB) is a method for extracting information from one random variable X that is relevant for predicting another random variable Y. To do so, IB identifies an intermediate \"bottleneck\" variable T that has low mutual information I(X;T) and high mutual information I(Y;T). The \"IB curve\" characterizes the set of bottleneck variables that achieve maximal I(Y;T) for a given I(X;T), and is typically explored by maximizing the \"IB Lagrangian\", I(Y;T) - βI(X;T). In some cases, Y is a deterministic function of X, including many classification problems in supervised learning where the output class Y is a deterministic function of the input X. We demonstrate three caveats when using IB in any situation where Y is a deterministic function of X: (1) the IB curve cannot be recovered by maximizing the IB Lagrangian for different values of β; (2) there are \"uninteresting\" trivial solutions at all points of the IB curve; and (3) for multi-layer classifiers that achieve low prediction error, different layers cannot exhibit a strict trade-off between compression and prediction, contrary to a recent proposal. We also show that when Y is a small perturbation away from being a deterministic function of X, these three caveats arise in an approximate way. To address problem (1), we propose a functional that, unlike the IB Lagrangian, can recover the IB curve in all cases. We demonstrate the three caveats on the MNIST dataset.", "target": ["L'information bottleneck si comporta in modi sorprendenti ogni volta che l'output è una funzione deterministica dell'input.", "Sostiene che la maggior parte dei problemi reali di classificazione mostra una relazione deterministica tra le label di classe e gli input X ed esplora diversi problemi che derivano da tali patologie.", "Esplora i problemi che sorgono quando si applicano i concetti di information bottleneck ai modelli deterministici di apprendimento supervisionato", "Gli autori chiariscono diversi comportamenti controintuitivi del metodo dell'information bottleneck per l'apprendimento supervisionato di una regola deterministica."]} +{"source": "We prove, under two sufficient conditions, that idealised models can have no adversarial examples. We discuss which idealised models satisfy our conditions, and show that idealised Bayesian neural networks (BNNs) satisfy these. We continue by studying near-idealised BNNs using HMC inference, demonstrating the theoretical ideas in practice. We experiment with HMC on synthetic data derived from MNIST for which we know the ground-truth image density, showing that near-perfect epistemic uncertainty correlates to density under image manifold, and that adversarial images lie off the manifold in our setting. This suggests why MC dropout, which can be seen as performing approximate inference, has been observed to be an effective defence against adversarial examples in practice; We highlight failure-cases of non-idealised BNNs relying on dropout, suggesting a new attack for dropout models and a new defence as well. Lastly, we demonstrate the defence on a cats-vs-dogs image classification task with a VGG13 variant.", "target": ["Dimostriamo che le reti neurali bayesiane idealizzate non possono avere adversarial example, e diamo prove empiriche con BNN del mondo reale.", "L'articolo studia l'adversarial robustness dei classificatori bayesiani e stabilisce due condizioni che dimostrano essere provatamente sufficienti per \"modelli idealizzati\" su \"dataset idealizzati\" per non avere adversarial example", "Il paper propone una classe di classificatori discriminativi bayesiani che non hanno adversarial example."]} +{"source": "Deep neural networks are susceptible to adversarial attacks. In computer vision, well-crafted perturbations to images can cause neural networks to make mistakes such as confusing a cat with a computer. Previous adversarial attacks have been designed to degrade performance of models or cause machine learning models to produce specific outputs chosen ahead of time by the attacker. We introduce attacks that instead reprogram the target model to perform a task chosen by the attacker without the attacker needing to specify or compute the desired output for each test-time input. This attack finds a single adversarial perturbation, that can be added to all test-time inputs to a machine learning model in order to cause the model to perform a task chosen by the adversary—even if the model was not trained to do this task. These perturbations can thus be considered a program for the new task. We demonstrate adversarial reprogramming on six ImageNet classification models, repurposing these models to perform a counting task, as well as classification tasks: classification of MNIST and CIFAR-10 examples presented as inputs to the ImageNet model.", "target": ["Introduciamo la prima istanza di adversarial attack che riprogramma il modello bersaglio per eseguire un task scelto dall'attaccante senza che l'attaccante abbia bisogno di specificare o calcolare l'output desiderato per ogni input al test-time.", "Gli autori presentano un nuovo schema di adversarial attack in cui una rete neurale viene riprogrammata per svolgere un task diverso da quello su cui è stata originariamente addestrata", "Questo articolo ha proposto la \"riprogrammazione adversarial\" di reti neurali ben addestrate e fissate e mostra che la riprogrammazione adversarial è meno efficace sulle reti non addestrate.", "L'articolo estende l'idea degli \"adversarial attack\" nell'apprendimento supervisionato delle NN a una completa riprogrammazione della soluzione di una rete addestrata."]} +{"source": "As shown in recent research, deep neural networks can perfectly fit randomly labeled data, but with very poor accuracy on held out data. This phenomenon indicates that loss functions such as cross-entropy are not a reliable indicator of generalization. This leads to the crucial question of how generalization gap should be predicted from the training data and network parameters. In this paper, we propose such a measure, and conduct extensive empirical studies on how well it can predict the generalization gap. Our measure is based on the concept of margin distribution, which are the distances of training points to the decision boundary. We find that it is necessary to use margin distributions at multiple layers of a deep network. On the CIFAR-10 and the CIFAR-100 datasets, our proposed measure correlates very strongly with the generalization gap. In addition, we find the following other factors to be of importance: normalizing margin values for scale independence, using characterizations of margin distribution rather than just the margin (closest distance to decision boundary), and working in log space instead of linear space (effectively using a product of margins rather than a sum). Our measure can be easily applied to feedforward deep networks with any architecture and may point towards new training loss functions that could enable better generalization.", "target": ["Sviluppiamo un nuovo schema per prevedere il gap di generalizzazione nelle deep network con alta precisione.", "Gli autori suggeriscono di usare un margine geometrico e una distribuzione dei margini layer-wise per prevedere il generalization gap.", "Mostra empiricamente un'interessante connessione tra le statistiche di margine proposte e il gap di generalizzazione, che può essere utilizzato per fornire alcune intuizioni prescrittive verso la comprensione della generalizzazione nelle deep neural network."]} +{"source": "We propose a new algorithm to learn a one-hidden-layer convolutional neural network where both the convolutional weights and the outputs weights are parameters to be learned. Our algorithm works for a general class of (potentially overlapping) patches, including commonly used structures for computer vision tasks. Our algorithm draws ideas from (1) isotonic regression for learning neural networks and (2) landscape analysis of non-convex matrix factorization problems. We believe these findings may inspire further development in designing provable algorithms for learning neural networks and other complex models. While our focus is theoretical, we also present experiments that illustrate our theoretical findings.", "target": ["Proponiamo un algoritmo per recuperare in modo dimostrabile i parametri (pesi di convoluzione e di output) di una rete convoluzionale con patch sovrapposte.", "Questo articolo studia l'apprendimento teorico delle reti neurali convoluzionali ad un hidden layer, risultando in un algoritmo di apprendimento e garanzie dimostrabili utilizzando l'algoritmo.", "Questo articolo fornisce un nuovo algoritmo per l'apprendimento di una rete neurale a due layer che coinvolge un singolo filtro convoluzionale e un vettore di peso per diverse posizioni."]} +{"source": "Since their invention, generative adversarial networks (GANs) have become a popular approach for learning to model a distribution of real (unlabeled) data. Convergence problems during training are overcome by Wasserstein GANs which minimize the distance between the model and the empirical distribution in terms of a different metric, but thereby introduce a Lipschitz constraint into the optimization problem. A simple way to enforce the Lipschitz constraint on the class of functions, which can be modeled by the neural network, is weight clipping. Augmenting the loss by a regularization term that penalizes the deviation of the gradient norm of the critic (as a function of the network's input) from one, was proposed as an alternative that improves training. We present theoretical arguments why using a weaker regularization term enforcing the Lipschitz constraint is preferable. These arguments are supported by experimental results on several data sets.", "target": ["Un nuovo termine di regolarizzazione può migliorare l'allenamento delle Wasserstein GAN", "L'articolo propone uno schema di regolarizzazione per Wasserstein GAN basato sul rilassamento dei vincoli sulla costante di Lipschitz pari ad 1.", "L'articolo tratta della regolarizzazione/penalizzazione nel fitting delle GAN, quando si basa su una metrica L_1 di Wasserstein."]} +{"source": "We introduce a new method for training GANs by applying the Wasserstein-2 metric proximal on the generators. The approach is based on the gradient operator induced by optimal transport, which connects the geometry of sample space and parameter space in implicit deep generative models. From this theory, we obtain an easy-to-implement regularizer for the parameter updates. Our experiments demonstrate that this method improves the speed and stability in training GANs in terms of wall-clock time and Fr\\'echet Inception Distance (FID) learning curves.", "target": ["Proponiamo il metodo prossimale di Wasserstein per il training delle GAN.", "Propone una nuova procedura per le GAN che tiene conto dei punti generati nell'iterazione precedente e aggiorna il generatore per l volte.", "Considera l'apprendimento a gradiente naturale nel learning delle GAN, dove viene impiegata la struttura Riemanniana indotta dalla distanza Wasserstein-2.", "L'articolo intende utilizzare il gradiente naturale indotto dalla distanza Wasserstein-2 per addestrare il generatore nelle GAN e gli autori propongono l'operatore prossimale di Wasserstein come regolarizzazione."]} +{"source": "Influence diagrams provide a modeling and inference framework for sequential decision problems, representing the probabilistic knowledge by a Bayesian network and the preferences of an agent by utility functions over the random variables and decision variables. MDPs and POMDPS, widely used for planning under uncertainty can also be represented by influence diagrams. The time and space complexity of computing the maximum expected utility (MEU) and its maximizing policy is exponential in the induced width of the underlying graphical model, which is often prohibitively large due to the growth of the information set under the sequence of decisions. In this paper, we develop a weighted mini-bucket approach for bounding the MEU. These bounds can be used as a stand-alone approximation that can be improved as a function of a controlling i-bound parameter . They can also be used as heuristic functions to guide search, especially for planning such as MDPs and POMDPs. We evaluate the scheme empirically against state-of-the-art, thus illustrating its potential.", "target": ["Questo articolo introduce una funzione euristica basata sull'eliminazione per il processo decisionale sequenziale, adatta a guidare algoritmi di ricerca AND/OR per risolvere diagrammi di influenza.", "Generalizza l'euristica di inferenza dei minibucket ai diagrammi di influenza."]} +{"source": "Probabilistic Neural Networks deal with various sources of stochasticity: input noise, dropout, stochastic neurons, parameter uncertainties modeled as random variables, etc. In this paper we revisit a feed-forward propagation approach that allows one to estimate for each neuron its mean and variance w.r.t. all mentioned sources of stochasticity. In contrast, standard NNs propagate only point estimates, discarding the uncertainty. Methods propagating also the variance have been proposed by several authors in different context. The view presented here attempts to clarify the assumptions and derivation behind such methods, relate them to classical NNs and broaden their scope of applicability. The main technical contributions are new approximations for the distributions of argmax and max-related transforms, which allow for fully analytic uncertainty propagation in networks with softmax and max-pooling layers as well as leaky ReLU activations. We evaluate the accuracy of the approximation and suggest a simple calibration. Applying the method to networks with dropout allows for faster training and gives improved test likelihoods without the need of sampling.", "target": ["Approssimazione della media e della varianza dell'output di una NN nel caso di input rumorosi / dropout / parametri incerti. Approssimazioni analitiche per i layer argmax, softmax e max.", "Gli autori si concentrano sul problema della propagazione dell'incertezza nelle DNN", "Questo articolo rivisita la propagazione feed-forward della media e della varianza nei neuroni, affrontando il problema della propagazione dell'incertezza attraverso layer di max-pooling e softmax."]} +{"source": "Generative Adversarial Networks are one of the leading tools in generative modeling, image editing and content creation. However, they are hard to train as they require a delicate balancing act between two deep networks fighting a never ending duel. Some of the most promising adversarial models today minimize a Wasserstein objective. It is smoother and more stable to optimize. In this paper, we show that the Wasserstein distance is just one out of a large family of objective functions that yield these properties. By making the discriminator of a GAN robust to adversarial attacks we can turn any GAN objective into a smooth and stable loss. We experimentally show that any GAN objective, including Wasserstein GANs, benefit from adversarial robustness both quantitatively and qualitatively. The training additionally becomes more robust to suboptimal choices of hyperparameters, model architectures, or objective functions.", "target": ["Un discriminatore che non è facilmente ingannato dall'adversarial example rende il training GAN più robusto e porta ad un obiettivo più smooth.", "Questo articolo propone un nuovo modo per stabilizzare il processo di training delle GAN regolarizzando il discriminatore per essere robusto agli adversarial example.", "L'articolo propone un modo sistematico di addestrare le GAN con termini di regolarizzazione della robustezza, permettendo un training più fluido.", "Presenta l'idea che rendendo un discriminatore robusto alle adversarial perturbation l'obiettivo GAN può essere reso smooth, il che si traduce in risultati migliori sia visivamente che in termini di FID."]} +{"source": "We propose a method to learn stochastic activation functions for use in probabilistic neural networks. First, we develop a framework to embed stochastic activation functions based on Gaussian processes in probabilistic neural networks. Second, we analytically derive expressions for the propagation of means and covariances in such a network, thus allowing for an efficient implementation and training without the need for sampling. Third, we show how to apply variational Bayesian inference to regularize and efficiently train this model. The resulting model can deal with uncertain inputs and implicitly provides an estimate of the confidence of its predictions. Like a conventional neural network it can scale to datasets of arbitrary size and be extended with convolutional and recurrent connections, if desired.", "target": ["Modelliamo la funzione di attivazione di ogni neurone come un processo gaussiano e la impariamo insieme al peso con l'inferenza variazionale.", "Propone di mettere prior di processo gaussiano sulla forma funzionale di ogni funzione di attivazione nella rete neurale per imparare la forma delle funzioni di attivazione."]} +{"source": "Recent results from linear algebra stating that any matrix can be decomposed into products of diagonal and circulant matrices has lead to the design of compact deep neural network architectures that perform well in practice. In this paper, we bridge the gap between these good empirical results and the theoretical approximation capabilities of Deep diagonal-circulant ReLU networks. More precisely, we first demonstrate that a Deep diagonal-circulant ReLU networks of bounded width and small depth can approximate a deep ReLU network in which the dense matrices are of low rank. Based on this result, we provide new bounds on the expressive power and universal approximativeness of this type of networks. We support our experimental results with thorough experiments on a large, real world video classification problem.", "target": ["Forniamo uno studio teorico delle proprietà delle reti ReLU circolanti-diagonali profonde e dimostriamo che sono approssimatori universali a larghezza limitata.", "L'articolo propone di utilizzare matrici circolanti e diagonali per accelerare il calcolo e ridurre i requisiti di memoria nelle reti neurali.", "Questo articolo dimostra che le reti ReLU diagonali-circolari di larghezza limitata (DC-ReLU) sono approssimatori universali."]} +{"source": "Camera drones, a rapidly emerging technology, offer people the ability to remotely inspect an environment with a high degree of mobility and agility. However, manual remote piloting of a drone is prone to errors. In contrast, autopilot systems can require a significant degree of environmental knowledge and are not necessarily designed to support flexible visual inspections. Inspired by camera manipulation techniques in interactive graphics, we designed StarHopper, a novel touch screen interface for efficient object-centric camera drone navigation, in which a user directly specifies the navigation of a drone camera relative to a specified object of interest. The system relies on minimal environmental information and combines both manual and automated control mechanisms to give users the freedom to remotely explore an environment with efficiency and accuracy. A lab study shows that StarHopper offers an efficiency gain of 35.4% over manual piloting, complimented by an overall user preference towards our object-centric navigation system.", "target": ["StarHopper è una nuova interfaccia touch screen per la navigazione object-centric, efficiente e flessibile dei droni con telecamera", "Gli autori delineano una nuova interfaccia di controllo del drone chiamata StarHopper, combinando il pilotaggio automatico e manuale in una nuova interfaccia di navigazione ibrida che si sbarazza del presupposto che l'oggetto di destinazione è già nel FOV del drone utilizzando una telecamera supplementare sopra la testa.", "Questo articolo presenta StarHopper, un sistema per la navigazione semiautomatica dei droni nel contesto dell'ispezione remota.", "Presenta StarHopper, un'applicazione che utilizza tecniche di computer vision con input tattile per supportare il pilotaggio dei droni con un approccio centrato sull'oggetto."]} +{"source": "Recurrent neural networks (RNN), convolutional neural networks (CNN) and self-attention networks (SAN) are commonly used to produce context-aware representations. RNN can capture long-range dependency but is hard to parallelize and not time-efficient. CNN focuses on local dependency but does not perform well on some tasks. SAN can model both such dependencies via highly parallelizable computation, but memory requirement grows rapidly in line with sequence length. In this paper, we propose a model, called \"bi-directional block self-attention network (Bi-BloSAN)\", for RNN/CNN-free sequence encoding. It requires as little memory as RNN but with all the merits of SAN. Bi-BloSAN splits the entire sequence into blocks, and applies an intra-block SAN to each block for modeling local context, then applies an inter-block SAN to the outputs for all blocks to capture long-range dependency. Thus, each SAN only needs to process a short sequence, and only a small amount of memory is required. Additionally, we use feature-level attention to handle the variation of contexts around the same word, and use forward/backward masks to encode temporal order information. On nine benchmark datasets for different NLP tasks, Bi-BloSAN achieves or improves upon state-of-the-art accuracy, and shows better efficiency-memory trade-off than existing RNN/CNN/SAN.", "target": ["Una rete di self-attention per la codifica di sequenze senza RNN/CNN con un piccolo consumo di memoria, calcolo altamente parallelizzabile e prestazioni allo stato dell'arte su diversi task NLP", "Propone di applicare self-attention a due livelli per limitare la richiesta di memoria nei modelli basati sull'attention con un impatto trascurabile sulla velocità.", "Questo articolo introduce il modello di self-attention bidirezionale a blocchi come un encoder generale per vari task di modellazione di sequenze in NLP"]} +{"source": "End-to-end neural models have made significant progress in question answering, however recent studies show that these models implicitly assume that the answer and evidence appear close together in a single document. In this work, we propose the Coarse-grain Fine-grain Coattention Network (CFC), a new question answering model that combines information from evidence across multiple documents. The CFC consists of a coarse-grain module that interprets documents with respect to the query then finds a relevant answer, and a fine-grain module which scores each candidate answer by comparing its occurrences across all of the documents with the query. We design these modules using hierarchies of coattention and self-attention, which learn to emphasize different parts of the input. On the Qangaroo WikiHop multi-evidence question answering task, the CFC obtains a new state-of-the-art result of 70.6% on the blind test set, outperforming the previous best by 3% accuracy despite not using pretrained contextual encoders.", "target": ["Un nuovo modello allo stato dell'arte per la question answering multi-evidenza utilizzando l'attention gerarchica a grana grossa e fine.", "Propone un metodo per il QA multi-hop basato su due moduli separati (moduli coarse-grained e fine-grained).", "Questo articolo propone un'interessante architettura di rete di co-attention coarse-grained e fine-grained per affrontare question answering multi-evidence", "Si concentra su QA multi-choice e propone un framework di punteggio coarse-to-fine."]} +{"source": "Generative adversarial networks (GANs) are an expressive class of neural generative models with tremendous success in modeling high-dimensional continuous measures. In this paper, we present a scalable method for unbalanced optimal transport (OT) based on the generative-adversarial framework. We formulate unbalanced OT as a problem of simultaneously learning a transport map and a scaling factor that push a source measure to a target measure in a cost-optimal manner. We provide theoretical justification for this formulation, showing that it is closely related to an existing static formulation by Liero et al. (2018). We then propose an algorithm for solving this problem based on stochastic alternating gradient updates, similar in practice to GANs, and perform numerical experiments demonstrating how this methodology can be applied to population modeling.", "target": ["Proponiamo una nuova metodologia per il trasporto ottimale sbilanciato utilizzando generative adversarial network.", "Gli autori considerano il problema del trasporto ottimale sbilanciato tra due misure con massa totale diversa usando un algoritmo stocastico min-max e uno scaling locale", "Gli autori propongono un approccio per stimare il trasporto ottimale sbilanciato tra misure campionate che scala bene nella dimensione e nel numero di sample.", "L'articolo introduce una formulazione statica per il trasporto ottimale sbilanciato imparando simultaneamente una mappa di trasporto T e un fattore di scala xi."]} +{"source": "Extracting saliency maps, which indicate parts of the image important to classification, requires many tricks to achieve satisfactory performance when using classifier-dependent methods. Instead, we propose classifier-agnostic saliency map extraction, which finds all parts of the image that any classifier could use, not just one given in advance. We observe that the proposed approach extracts higher quality saliency maps and outperforms existing weakly-supervised localization techniques, setting the new state of the art result on the ImageNet dataset.", "target": ["Proponiamo un nuovo metodo di estrazione delle mappe di saliency che risulta nell'estrazione di mappe di qualità superiore.", "Propone un metodo indipendente dal classificatore per l'estrazione delle mappe di saliency.", "Questo articolo introduce un nuovo estrattore di mappe di saliency che sembra migliorare i risultati allo stato dell'arte.", "Gli autori sostengono che quando una mappa di saliency estratta dipende direttamente da un modello, allora potrebbe non essere utile per un classificatore diverso, e suggerisce uno schema per approssimare la soluzione."]} +{"source": "Unsupervised image-to-image translation has gained considerable attention due to the recent impressive progress based on generative adversarial networks (GANs). However, previous methods often fail in challenging cases, in particular, when an image has multiple target instances and a translation task involves significant changes in shape, e.g., translating pants to skirts in fashion images. To tackle the issues, we propose a novel method, coined instance-aware GAN (InstaGAN), that incorporates the instance information (e.g., object segmentation masks) and improves multi-instance transfiguration. The proposed method translates both an image and the corresponding set of instance attributes while maintaining the permutation invariance property of the instances. To this end, we introduce a context preserving loss that encourages the network to learn the identity function outside of target instances. We also propose a sequential mini-batch inference/training technique that handles multiple instances with a limited GPU memory and enhances the network to generalize better for multiple instances. Our comparative evaluation demonstrates the effectiveness of the proposed method on different image datasets, in particular, in the aforementioned challenging cases. Code and results are available in https://github.com/sangwoomo/instagan", "target": ["Proponiamo un nuovo metodo per incorporare l'insieme degli attributi di istanza per la traduzione image-to-image.", "Questo articolo propone un metodo - InstaGAN - che si basa su CycleGAN prendendo in considerazione le informazioni dell'istanza sotto forma di maschere di segmentazione per istanza, con risultati che si confrontano favorevolmente con CycleGAN e altre baseline.", "Propone di aggiungere maschere di segmentazione instance-aware per il problema della traduzione image-to-image non abbinata."]} +{"source": "Deep neural networks (DNNs) generalize remarkably well without explicit regularization even in the strongly over-parametrized regime where classical learning theory would instead predict that they would severely overfit. While many proposals for some kind of implicit regularization have been made to rationalise this success, there is no consensus for the fundamental reason why DNNs do not strongly overfit. In this paper, we provide a new explanation. By applying a very general probability-complexity bound recently derived from algorithmic information theory (AIT), we argue that the parameter-function map of many DNNs should be exponentially biased towards simple functions. We then provide clear evidence for this strong simplicity bias in a model DNN for Boolean functions, as well as in much larger fully connected and convolutional networks trained on CIFAR10 and MNIST. As the target functions in many real problems are expected to be highly structured, this intrinsic simplicity bias helps explain why deep networks generalize well on real world problems. This picture also facilitates a novel PAC-Bayes approach where the prior is taken over the DNN input-output function space, rather than the more conventional prior over parameter space. If we assume that the training algorithm samples parameters close to uniformly within the zero-error region then the PAC-Bayes theorem can be used to guarantee good expected generalization for target functions producing high-likelihood training sets. By exploiting recently discovered connections between DNNs and Gaussian processes to estimate the marginal likelihood, we produce relatively tight generalization PAC-Bayes error bounds which correlate well with the true error on realistic datasets such as MNIST and CIFAR10 and for architectures including convolutional and fully connected networks.", "target": ["La mappa parametro-funzione delle deep network è enormemente biased; questo può spiegare perché generalizzano. Usiamo PAC-Bayes e processi gaussiani per ottenere bound non banali.", "L'articolo studia le capacità di generalizzazione delle deep neural network, con l'aiuto della teoria dell'apprendimento PAC-Bayesiana e delle intuizioni empiricamente sostenute.", "Questo articolo propone una spiegazione dei comportamenti di generalizzazione delle grandi reti neurali iper-parametrizzate sostenendo che la mappa parametro-funzione nelle reti neurali è biased verso funzioni \"semplici\" e il comportamento di generalizzazione sarà buono se il concetto target è anche \"semplice\"."]} +{"source": "We establish a theoretical link between evolutionary algorithms and variational parameter optimization of probabilistic generative models with binary hidden variables. While the novel approach is independent of the actual generative model, here we use two such models to investigate its applicability and scalability: a noisy-OR Bayes Net (as a standard example of binary data) and Binary Sparse Coding (as a model for continuous data). Learning of probabilistic generative models is first formulated as approximate maximum likelihood optimization using variational expectation maximization (EM). We choose truncated posteriors as variational distributions in which discrete latent states serve as variational parameters. In the variational E-step, the latent states are then optimized according to a tractable free-energy objective . Given a data point, we can show that evolutionary algorithms can be used for the variational optimization loop by (A)~considering the bit-vectors of the latent states as genomes of individuals, and by (B)~defining the fitness of the individuals as the (log) joint probabilities given by the used generative model. As a proof of concept, we apply the novel evolutionary EM approach to the optimization of the parameters of noisy-OR Bayes nets and binary sparse coding on artificial and real data (natural image patches). Using point mutations and single-point cross-over for the evolutionary algorithm, we find that scalable variational EM algorithms are obtained which efficiently improve the data likelihood. In general we believe that, with the link established here, standard as well as recent results in the field of evolutionary optimization can be leveraged to address the difficult problem of parameter optimization in generative models.", "target": ["Presentiamo Evolutionary EM come un nuovo algoritmo per il training unsupervised di modelli generativi con variabili latenti binarie che collega intimamente l'EM variazionale con l'ottimizzazione evolutiva", "L'articolo presenta una combinazione di calcolo evolutivo ed EM variazionale per modelli con variabili latenti binarie rappresentate tramite un'approssimazione particle-based", "L'articolo fa un tentativo di integrare strettamente gli algoritmi di training di expectation-maximization con gli algoritmi evolutivi."]} +{"source": "While deep neural networks have achieved groundbreaking prediction results in many tasks, there is a class of data where existing architectures are not optimal -- sequences of probability distributions. Performing forward prediction on sequences of distributions has many important applications. However, there are two main challenges in designing a network model for this task. First, neural networks are unable to encode distributions compactly as each node encodes just a real value. A recent work of Distribution Regression Network (DRN) solved this problem with a novel network that encodes an entire distribution in a single node, resulting in improved accuracies while using much fewer parameters than neural networks. However, despite its compact distribution representation, DRN does not address the second challenge, which is the need to model time dependencies in a sequence of distributions. In this paper, we propose our Recurrent Distribution Regression Network (RDRN) which adopts a recurrent architecture for DRN. The combination of compact distribution representation and shared weights architecture across time steps makes RDRN suitable for modeling the time dependencies in a distribution sequence. Compared to neural networks and DRN, RDRN achieves the best prediction performance while keeping the network compact.", "target": ["Proponiamo un modello efficiente di rete ricorrente per la forward prediction su distribuzioni variabili nel tempo.", "Questo articolo propone un metodo per creare reti neurali che mappano le distribuzioni storiche sulle distribuzioni e applica il metodo a diversi task di predizione della distribuzione.", "Propone una Recurent Distribution Regression Network che utilizza un'architettura ricorrente su un modello precedente di Distribution Regression Network.", "Questo articolo riguarda la regressione su distribuzioni di probabilità studiando distribuzioni variabili nel tempo in una rete neurale ricorrente"]} +{"source": "We present graph attention networks (GATs), novel neural network architectures that operate on graph-structured data, leveraging masked self-attentional layers to address the shortcomings of prior methods based on graph convolutions or their approximations. By stacking layers in which nodes are able to attend over their neighborhoods' features, we enable (implicitly) specifying different weights to different nodes in a neighborhood, without requiring any kind of computationally intensive matrix operation (such as inversion) or depending on knowing the graph structure upfront. In this way, we address several key challenges of spectral-based graph neural networks simultaneously, and make our model readily applicable to inductive as well as transductive problems. Our GAT models have achieved or matched state-of-the-art results across four established transductive and inductive graph benchmarks: the Cora, Citeseer and Pubmed citation network datasets, as well as a protein-protein interaction dataset (wherein test graphs remain unseen during training).", "target": ["Un nuovo approccio all'elaborazione dei dati strutturati a grafo da parte delle reti neurali, sfruttando l'attention sul neighborhood di un nodo. Raggiunge risultati allo stato dell'arte su task di citation network trasduttiva e un task di interazione proteina-proteina induttiva.", "Questo articolo propone un nuovo metodo per classificare i nodi di un grafo, che può essere utilizzato in scenari semi-supervisionati e su un grafo completamente nuovo.", "L'articolo introduce un'architettura di rete neurale per operare su dati strutturati a grafo chiamata Graph Attention Network.", "Fornisce una discussione corretta e quasi completa dello stato dell'arte degli approcci all'apprendimento di rappresentazioni vettoriali per i nodi di un grafo."]} +{"source": "While bigger and deeper neural network architectures continue to advance the state-of-the-art for many computer vision tasks, real-world adoption of these networks is impeded by hardware and speed constraints. Conventional model compression methods attempt to address this problem by modifying the architecture manually or using pre-defined heuristics. Since the space of all reduced architectures is very large, modifying the architecture of a deep neural network in this way is a difficult task. In this paper, we tackle this issue by introducing a principled method for learning reduced network architectures in a data-driven way using reinforcement learning. Our approach takes a larger 'teacher' network as input and outputs a compressed 'student' network derived from the 'teacher' network. In the first stage of our method, a recurrent policy network aggressively removes layers from the large 'teacher' model. In the second stage, another recurrent policy network carefully reduces the size of each remaining layer. The resulting network is then evaluated to obtain a reward -- a score based on the accuracy and compression of the network. Our approach uses this reward signal with policy gradients to train the policies to find a locally optimal student network. Our experiments show that we can achieve compression rates of more than 10x for models such as ResNet-34 while maintaining similar performance to the input 'teacher' network. We also present a valuable transfer learning result which shows that policies which are pre-trained on smaller 'teacher' networks can be used to rapidly speed up training on larger 'teacher' networks.", "target": ["Un nuovo approccio basato sul reinforcement learning per comprimere le deep neural network con la knowledge distillation", "Questo articolo propone di usare il reinforcement learning invece di euristiche predefinite per determinare la struttura del modello compresso nel processo di knowledge distillation", "Introduce un metodo per la compressione da rete a rete, che utilizza i policy gradient per ottimizzare due policy che comprimono un teacher model forte in uno student model forte ma più piccolo."]} +{"source": "Recent advances in conditional image generation tasks, such as image-to-image translation and image inpainting, are largely accounted to the success of conditional GAN models, which are often optimized by the joint use of the GAN loss with the reconstruction loss. However, we reveal that this training recipe shared by almost all existing methods causes one critical side effect: lack of diversity in output samples. In order to accomplish both training stability and multimodal output generation, we propose novel training schemes with a new set of losses named moment reconstruction losses that simply replace the reconstruction loss. We show that our approach is applicable to any conditional generation tasks by performing thorough experiments on image-to-image translation, super-resolution and image inpainting using Cityscapes and CelebA dataset. Quantitative evaluations also confirm that our methods achieve a great diversity in outputs while retaining or even improving the visual fidelity of generated samples.", "target": ["Dimostriamo che il mode collapse nelle GAN condizionali è in gran parte attribuito a una mancata corrispondenza tra la loss di ricostruzione e la loss GAN e introduciamo una serie di nuove funzioni di loss come alternative alla loss di ricostruzione.", "L'articolo propone una modifica al tradizionale obiettivo GAN condizionale per promuovere una generazione di immagini diverse e multimodali.", "Questo articolo propone un'alternativa agli errori L1/L2 che sono usati per aumentare le loss adversarial quando si addestrano le GAN condizionali."]} +{"source": "Generative models are important tools to capture and investigate the properties of complex empirical data. Recent developments such as Generative Adversarial Networks (GANs) and Variational Auto-Encoders (VAEs) use two very similar, but \\textit{reverse}, deep convolutional architectures, one to generate and one to extract information from data. Does learning the parameters of both architectures obey the same rules? We exploit the causality principle of independence of mechanisms to quantify how the weights of successive layers adapt to each other. Using the recently introduced Spectral Independence Criterion, we quantify the dependencies between the kernels of successive convolutional layers and show that those are more independent for the generative process than for information extraction, in line with results from the field of causal inference. In addition, our experiments on generation of human faces suggest that more independence between successive layers of generators results in improved performance of these architectures.", "target": ["Usiamo l'inferenza causale per caratterizzare l'architettura dei modelli generativi", "Questo articolo esamina la natura dei filtri convoluzionali nell'encoder e nel decoder di un VAE, e nel generatore e nel discriminatore di una GAN.", "Questo lavoro sfrutta il principio di causalità per quantificare come i pesi dei layer successivi si adattino gli uni agli altri."]} +{"source": "Many deep reinforcement learning approaches use graphical state representations, this means visually distinct games that share the same underlying structure cannot effectively share knowledge. This paper outlines a new approach for learning underlying game state embeddings irrespective of the visual rendering of the game state. We utilise approaches from multi-task learning and domain adaption in order to place visually distinct game states on a shared embedding manifold. We present our results in the context of deep reinforcement learning agents.", "target": ["Un approccio all'apprendimento di uno spazio di embedding condiviso tra giochi visivamente distinti.", "Un nuovo approccio per l'apprendimento della struttura di base dei giochi visivamente distinti che combina layer convoluzionali per l'elaborazione delle immagini di input, Asynchronous Advantage Actor Critic per deep reinforcement learning e l'approccio adversarial per forzare la rappresentazione di embedding ad essere indipendente dalla rappresentazione visiva dei giochi", "Introduce un metodo per imparare una policy su giochi visivamente distinti adattando il deep reinforcement learning.", "Questo articolo discute un'architettura di un agente che usa una rappresentazione condivisa per addestrarsi su task multipli con diverse statistiche visive a livello di sprite"]} +{"source": "We study discrete time dynamical systems governed by the state equation $h_{t+1}=ϕ(Ah_t+Bu_t)$. Here A,B are weight matrices, ϕ is an activation function, and $u_t$ is the input data. This relation is the backbone of recurrent neural networks (e.g. LSTMs) which have broad applications in sequential learning tasks. We utilize stochastic gradient descent to learn the weight matrices from a finite input/state trajectory $(u_t,h_t)_{t=0}^N$. We prove that SGD estimate linearly converges to the ground truth weights while using near-optimal sample size. Our results apply to increasing activations whose derivatives are bounded away from zero. The analysis is based on i) an SGD convergence result with nonlinear activations and ii) careful statistical characterization of the state vector. Numerical experiments verify the fast convergence of SGD on ReLU and leaky ReLU in consistence with our theory.", "target": ["Studiamo l'equazione di stato di una rete neurale ricorrente. Mostriamo che SGD può apprendere in modo efficiente la dinamica sconosciuta da poche osservazioni di input/output sotto opportune ipotesi.", "L'articolo studia sistemi dinamici a tempo discreto con un'equazione di stato non lineare, dimostrando che l'esecuzione di SGD su una traiettoria di lunghezza fissa dà una convergenza logaritmica.", "Questo lavoro considera il problema dell'apprendimento di un sistema dinamico non lineare in cui l'output è uguale allo stato.", "Questo articolo studia la capacità di SGD di apprendere le dinamiche di un sistema lineare e di un'attivazione non lineare."]} +{"source": "Although deep neural networks show their extraordinary power in various tasks, they are not feasible for deploying such large models on embedded systems due to high computational cost and storage space limitation. The recent work knowledge distillation (KD) aims at transferring model knowledge from a well-trained teacher model to a small and fast student model which can significantly help extending the usage of large deep neural networks on portable platform. In this paper, we show that, by properly defining the neuron manifold of deep neuron network (DNN), we can significantly improve the performance of student DNN networks through approximating neuron manifold of powerful teacher network. To make this, we propose several novel methods for learning neuron manifold from DNN model. Empowered with neuron manifold knowledge, our experiments show the great improvement across a variety of DNN architectures and training data. Compared with other KD methods, our Neuron Manifold Transfer (NMT) has best transfer ability of the learned features.", "target": ["Un nuovo metodo di knowledge distillation per transfer learning", "Il lavoro introduce un metodo di knowledge distillation utilizzando il concetto di manifold neuronale proposto.", "Propone un metodo di knowledge distillation in cui il manifold neurale è considerato come la conoscenza trasferita."]} +{"source": "We present a simple and general method to train a single neural network executable at different widths (number of channels in a layer), permitting instant and adaptive accuracy-efficiency trade-offs at runtime. Instead of training individual networks with different width configurations, we train a shared network with switchable batch normalization. At runtime, the network can adjust its width on the fly according to on-device benchmarks and resource constraints, rather than downloading and offloading different models. Our trained networks, named slimmable neural networks, achieve similar (and in many cases better) ImageNet classification accuracy than individually trained models of MobileNet v1, MobileNet v2, ShuffleNet and ResNet-50 at different widths respectively. We also demonstrate better performance of slimmable models compared with individual ones across a wide range of applications including COCO bounding-box object detection, instance segmentation and person keypoint detection without tuning hyper-parameters. Lastly we visualize and discuss the learned features of slimmable networks. Code and models are available at: https://github.com/JiahuiYu/slimmable_networks", "target": ["Presentiamo un metodo semplice e generale per addestrare una singola rete neurale eseguibile a diverse larghezze (numero di canali in un layer), permettendo trade-off di precisione-efficienza istantanei e adattivi al runtime.", "L'articolo propone un'idea di combinare insieme modelli di dimensioni diverse in una rete condivisa, migliorando notevolmente le prestazioni per il rilevamento", "Questo documento allena una singola rete eseguibile a diverse larghezze."]} +{"source": "Measuring visual (dis)similarity between two or more instances within a data distribution is a fundamental task in many applications, specially in image retrieval. Theoretically, non-metric distances are able to generate a more complex and accurate similarity model than metric distances, provided that the non-linear data distribution is precisely captured by the similarity model. In this work, we analyze a simple approach for deep learning networks to be used as an approximation of non-metric similarity functions and we study how these models generalize across different image retrieval datasets.", "target": ["Similarity network per imparare una stima di somiglianza visiva non metrica tra una coppia di immagini", "Gli autori propongono l'apprendimento di misure di somiglianza per la somiglianza visiva e ottengono con questo un miglioramento in dataset molto noti di Oxford e Parigi per image retrieval.", "L'articolo sostiene che è più adatto usare le distanze non metriche invece di quelle metriche."]} +{"source": "Training a model to perform a task typically requires a large amount of data from the domains in which the task will be applied. However, it is often the case that data are abundant in some domains but scarce in others. Domain adaptation deals with the challenge of adapting a model trained from a data-rich source domain to perform well in a data-poor target domain. In general, this requires learning plausible mappings between domains. CycleGAN is a powerful framework that efficiently learns to map inputs from one domain to another using adversarial training and a cycle-consistency constraint. However, the conventional approach of enforcing cycle-consistency via reconstruction may be overly restrictive in cases where one or more domains have limited training data. In this paper, we propose an augmented cyclic adversarial learning model that enforces the cycle-consistency constraint via an external task specific model, which encourages the preservation of task-relevant content as opposed to exact reconstruction. We explore digit classification in a low-resource setting in supervised, semi and unsupervised situation, as well as high resource unsupervised. In low-resource supervised setting, the results show that our approach improves absolute performance by 14% and 4% when adapting SVHN to MNIST and vice versa, respectively, which outperforms unsupervised domain adaptation methods that require high-resource unlabeled target domain. Moreover, using only few unsupervised target data, our approach can still outperforms many high-resource unsupervised models. Our model also outperforms on USPS to MNIST and synthetic digit to SVHN for high resource unsupervised adaptation. In speech domains, we similarly adopt a speech recognition model from each domain as the task specific model. Our approach improves absolute performance of speech recognition by 2% for female speakers in the TIMIT dataset, where the majority of training samples are from male voices.", "target": ["Un nuovo adversarial learning ciclico arricchito con un modello di task ausiliario che migliora le prestazioni di domain adaptation in setting supervisionati e non supervisionati con poche risorse ", "Propone un'estensione dei metodi di adattamento cycle-consistent adversarial al fine di affrontare la domain adaptation quando i dati target supervisionati disponibili sono limitati.", "Questo articolo introduce un approccio di domain adaptation basato sull'idea di GAN ciclica e propone due diversi algoritmi."]} +{"source": "Nodes residing in different parts of a graph can have similar structural roles within their local network topology. The identification of such roles provides key insight into the organization of networks and can also be used to inform machine learning on graphs. However, learning structural representations of nodes is a challenging unsupervised-learning task, which typically involves manually specifying and tailoring topological features for each node. Here we develop GraphWave, a method that represents each node’s local network neighborhood via a low-dimensional embedding by leveraging spectral graph wavelet diffusion patterns. We prove that nodes with similar local network neighborhoods will have similar GraphWave embeddings even though these nodes may reside in very different parts of the network. Our method scales linearly with the number of edges and does not require any hand-tailoring of topological features. We evaluate performance on both synthetic and real-world datasets, obtaining improvements of up to 71% over state-of-the-art baselines.", "target": ["Sviluppiamo un metodo per l'apprendimento delle firme strutturali nelle reti basato sulla diffusione delle wavelet spettrali del grafo.", "Si utilizzano i modelli di diffusione di wavelet spettrali del grafo del neighborhood locale di un nodo per fare l'embedding del nodo in uno spazio bidimensionale", "L'articolo ha derivato un modo per confrontare i nodi nel grafo basato sull'analisi wavelet del laplaciano del grafo."]} +{"source": "Driving simulators play an important role in vehicle research. However, existing virtual reality simulators do not give users a true sense of presence. UniNet is our driving simulator, designed to allow users to interact with and visualize simulated traffic in mixed reality. It is powered by SUMO and Unity. UniNet's modular architecture allows us to investigate interdisciplinary research topics such as vehicular ad-hoc networks, human-computer interaction, and traffic management. We accomplish this by giving users the ability to observe and interact with simulated traffic in a high fidelity driving simulator. We present a user study that subjectively measures user's sense of presence in UniNet. Our findings suggest that our novel mixed reality system does increase this sensation.", "target": ["Un simulatore di guida in realtà mista con telecamere stereo e VR passthrough valutato in uno studio utente con 24 partecipanti.", "Propone un sistema complicato per la simulazione di guida.", "Questo articolo presenta una configurazione di simulatore di guida in realtà mista per migliorare la sensazione di presenza", "Propone un simulatore di guida in realtà mista che incorpora la generazione di traffico e sostiene una \"presenza\" migliorata grazie a un sistema MR."]} +{"source": "We consider the problem of improving kernel approximation via feature maps. These maps arise as Monte Carlo approximation to integral representations of kernel functions and scale up kernel methods for larger datasets. We propose to use more efficient numerical integration technique to obtain better estimates of the integrals compared to the state-of-the-art methods. Our approach allows to use information about the integrand to enhance approximation and facilitates fast computations. We derive the convergence behavior and conduct an extensive empirical study that supports our hypothesis.", "target": ["Regole di quadratura per l'approssimazione del kernel.", "L'articolo propone di migliorare l'approssimazione del kernel delle feature casuali utilizzando regole di quadratura come le regole sferico-radiali stocastiche.", "Gli autori propongono una nuova versione dell'approccio random feature map per risolvere approssimativamente problemi di kernel su larga scala.", "Questo articolo mostra che le tecniche dovute a Genz & Monahan (1998) possono essere utilizzate per ottenere un basso errore di approssimazione del kernel nel framework della feature di Fourier casuali, un nuovo modo di applicare le regole di quadratura per migliorare l'approssimazione del kernel."]} +{"source": "Human world knowledge is both structured and flexible. When people see an object, they represent it not as a pixel array but as a meaningful arrangement of semantic parts. Moreover, when people refer to an object, they provide descriptions that are not merely true but also relevant in the current context. Here, we combine these two observations in order to learn fine-grained correspondences between language and contextually relevant geometric properties of 3D objects. To do this, we employed an interactive communication task with human participants to construct a large dataset containing natural utterances referring to 3D objects from ShapeNet in a wide variety of contexts. Using this dataset, we developed neural listener and speaker models with strong capacity for generalization. By performing targeted lesions of visual and linguistic input, we discovered that the neural listener depends heavily on part-related words and associates these words correctly with the corresponding geometric properties of objects, suggesting that it has learned task-relevant structure linking the two input modalities. We further show that a neural speaker that is `listener-aware' --- that plans its utterances according to how an imagined listener would interpret its words in context --- produces more discriminative referring expressions than an `listener-unaware' speaker, as measured by human performance in identifying the correct object.", "target": ["Come costruire parlanti/ascoltatori neurali che apprendono feature fine-grained di oggetti 3D, dal linguaggio referenziale.", "Gli autori forniscono uno studio sull'apprendimento dei riferimenti agli oggetti 3D, raccogliendo un dataset di espressioni referenziali e addestrando diversi modelli sperimentando una serie di scelte di architetture"]} +{"source": "Object-based factorizations provide a useful level of abstraction for interacting with the world. Building explicit object representations, however, often requires supervisory signals that are difficult to obtain in practice. We present a paradigm for learning object-centric representations for physical scene understanding without direct supervision of object properties. Our model, Object-Oriented Prediction and Planning (O2P2), jointly learns a perception function to map from image observations to object representations, a pairwise physics interaction function to predict the time evolution of a collection of objects, and a rendering function to map objects back to pixels. For evaluation, we consider not only the accuracy of the physical predictions of the model, but also its utility for downstream tasks that require an actionable representation of intuitive physics. After training our model on an image prediction task, we can use its learned representations to build block towers more complicated than those observed during training.", "target": ["Presentiamo un framework per l'apprendimento di rappresentazioni centrate sull'oggetto adatte alla pianificazione in task che richiedono una comprensione della fisica.", "L'articolo presenta una piattaforma per prevedere immagini di oggetti che interagiscono tra loro sotto l'effetto di forze gravitazionali.", "L'articolo presenta un metodo che impara a riprodurre \"block tower\" da un'immagine data.", "Propone un metodo che impara a ragionare sull'interazione fisica di diversi oggetti senza supervisione delle proprietà degli oggetti."]} +{"source": "We study the error landscape of deep linear and nonlinear neural networks with the squared error loss. Minimizing the loss of a deep linear neural network is a nonconvex problem, and despite recent progress, our understanding of this loss surface is still incomplete. For deep linear networks, we present necessary and sufficient conditions for a critical point of the risk function to be a global minimum. Surprisingly, our conditions provide an efficiently checkable test for global optimality, while such tests are typically intractable in nonconvex optimization. We further extend these results to deep nonlinear neural networks and prove similar sufficient conditions for global optimality, albeit in a more limited function space setting.", "target": ["Forniamo condizioni necessarie e sufficienti verificabili in modo efficiente per l'ottimalità globale nelle deep linear neural network, con alcune estensioni iniziali ai setting non lineari.", "L'articolo fornisce condizioni per l'ottimalità globale della loss delle deep linear neural network", "L'articolo fornisce risultati teorici sull'esistenza di minimi locali nella funzione obiettivo delle deep neural network.", "Studia alcune proprietà teoriche delle deep linear network."]} +{"source": "Recurrent auto-encoder model can summarise sequential data through an encoder structure into a fixed-length vector and then reconstruct into its original sequential form through the decoder structure. The summarised information can be used to represent time series features. In this paper, we propose relaxing the dimensionality of the decoder output so that it performs partial reconstruction. The fixed-length vector can therefore represent features only in the selected dimensions. In addition, we propose using rolling fixed window approach to generate samples. The change of time series features over time can be summarised as a smooth trajectory path. The fixed-length vectors are further analysed through additional visualisation and unsupervised clustering techniques. This proposed method can be applied in large-scale industrial processes for sensors signal analysis purpose where clusters of the vector representations can be used to reflect the operating states of selected aspects of the industrial system.", "target": ["Si utilizza il modello auto-encoder ricorrente per estrarre le feature multidimensionali delle serie temporali", "Questo articolo descrive un'applicazione di autoencoder ricorrente per analizzare serie temporali multidimensionali", "L'articolo descrive un modello di auto-encoder sequence to sequence che viene utilizzato per imparare rappresentazioni di sequenze, mostrando che, per la loro applicazione, si ottengono prestazioni migliori quando la rete viene addestrata solo per ricostruire un sottoinsieme delle misure dei dati.", "Propone una strategia ispirata al modello auto-encoder ricorrente in modo che il clustering di dati multidimensionali di serie temporali possa essere eseguito sulla base di vettori di contesto."]} +{"source": "We view molecule optimization as a graph-to-graph translation problem. The goal is to learn to map from one molecular graph to another with better properties based on an available corpus of paired molecules. Since molecules can be optimized in different ways, there are multiple viable translations for each input graph. A key challenge is therefore to model diverse translation outputs. Our primary contributions include a junction tree encoder-decoder for learning diverse graph translations along with a novel adversarial training method for aligning distributions of molecules. Diverse output distributions in our model are explicitly realized by low-dimensional latent vectors that modulate the translation process. We evaluate our model on multiple molecule optimization tasks and show that our model outperforms previous state-of-the-art baselines by a significant margin.", "target": ["Introduciamo un framework di encoder-decoder graph-to-graph per imparare diverse traduzioni di grafi.", "Propone un modello di traduzione graph-to-graph per l'ottimizzazione delle molecole ispirato all'analisi delle coppie molecolari abbinate.", "Estensione di JT-VAE nello scenario di traduzione graph-to-graph aggiungendo la variabile latente per catturare la multi-modalità e una regolarizzazione adversarial nello spazio latente", "Propone un sistema abbastanza complesso, che coinvolge molte scelte e componenti diverse, per ottenere composti chimici con proprietà migliorate a partire da un dato corpora."]} +{"source": "Partial differential equations (PDEs) are widely used across the physical and computational sciences. Decades of research and engineering went into designing fast iterative solution methods. Existing solvers are general purpose, but may be sub-optimal for specific classes of problems. In contrast to existing hand-crafted solutions, we propose an approach to learn a fast iterative solver tailored to a specific domain. We achieve this goal by learning to modify the updates of an existing solver using a deep neural network. Crucially, our approach is proven to preserve strong correctness and convergence guarantees. After training on a single geometry, our model generalizes to a wide variety of geometries and boundary conditions, and achieves 2-3 times speedup compared to state-of-the-art solvers.", "target": ["Impariamo un solutore neurale veloce per le PDE che ha garanzie di convergenza.", "Sviluppa un metodo per accelerare il metodo delle differenze finite nella risoluzione delle PDE e propone un framework rivisto per l'iterazione del punto fisso dopo la discretizzazione.", "Gli autori propongono un metodo lineare per accelerare i solutori di PDE."]} +{"source": "Variational Bayesian neural networks (BNN) perform variational inference over weights, but it is difficult to specify meaningful priors and approximating posteriors in a high-dimensional weight space. We introduce functional variational Bayesian neural networks (fBNNs), which maximize an Evidence Lower BOund (ELBO) defined directly on stochastic processes, i.e. distributions over functions. We prove that the KL divergence between stochastic processes is equal to the supremum of marginal KL divergences over all finite sets of inputs. Based on this, we introduce a practical training objective which approximates the functional ELBO using finite measurement sets and the spectral Stein gradient estimator. With fBNNs, we can specify priors which entail rich structure, including Gaussian processes and implicit stochastic processes. Empirically, we find that fBNNs extrapolate well using various structured priors, provide reliable uncertainty estimates, and can scale to large datasets.", "target": ["Eseguiamo l'inferenza funzionale variazionale sui processi stocastici definiti dalle reti neurali bayesiane.", "Adattamento di approssimazioni variazionali di reti neurali bayesiane in forma funzionale tenendo in considerazione la corrispondenza a un processo stocastico a priori implicitamente tramite sample.", "Presenta un nuovo obiettivo ELBO per il training delle BNN che permette di codificare nel modello prior più significativi piuttosto che i prior di peso meno informativi presenti in letteratura.", "Presenta un nuovo algoritmo di inferenza variazionale per modelli di reti neurali bayesiani in cui il prior è specificato in modo funzionale piuttosto che attraverso un prior sui pesi."]} +{"source": "Words are not created equal. In fact, they form an aristocratic graph with a latent hierarchical structure that the next generation of unsupervised learned word embeddings should reveal. In this paper, justified by the notion of delta-hyperbolicity or tree-likeliness of a space, we propose to embed words in a Cartesian product of hyperbolic spaces which we theoretically connect to the Gaussian word embeddings and their Fisher geometry. This connection allows us to introduce a novel principled hypernymy score for word embeddings. Moreover, we adapt the well-known Glove algorithm to learn unsupervised word embeddings in this type of Riemannian manifolds. We further explain how to solve the analogy task using the Riemannian parallel transport that generalizes vector arithmetics to this new type of geometry. Empirically, based on extensive experiments, we prove that our embeddings, trained unsupervised, are the first to simultaneously outperform strong and popular baselines on the tasks of similarity, analogy and hypernymy detection. In particular, for word hypernymy, we obtain new state-of-the-art on fully unsupervised WBLESS classification accuracy.", "target": ["Mappiamo le parole nello spazio iperbolico e facciamo la connessione con i word embedding gaussiani.", "Questo articolo adatta i word embedding di Glove a uno spazio iperbolico dato dal modello del semipiano di Poincare", "Questo articolo propone un approccio per implementare un modello di word embedding iperbolico basato su GLOVE, che è ottimizzato attraverso i metodi di ottimizzazione Riemanniana."]} +{"source": "Answering questions about a text frequently requires aggregating information from multiple places in that text. End-to-end neural network models, the dominant approach in the current literature, can theoretically learn how to distill and manipulate representations of the text without explicit supervision about how to do so. We investigate a canonical architecture for this task, the memory network, and analyze how effective it really is in the context of three multi-hop reasoning settings. In a simple synthetic setting, the path-finding task of the bAbI dataset, the model fails to learn the correct reasoning without additional supervision of its attention mechanism. However, with this supervision, it can perform well. On a real text dataset, WikiHop, the memory network gives nearly state-of-the-art performance, but does so without using its multi-hop capabilities. A tougher anonymized version of the WikiHop dataset is qualitatively similar to bAbI: the model fails to perform well unless it has additional supervision. We hypothesize that many \"multi-hop\" architectures do not truly learn this reasoning as advertised, though they could learn this reasoning if appropriately supervised.", "target": ["Le memory network non imparano il ragionamento multi-hop a meno che non le supervisioniamo.", "Il paper dichiara che il ragionamento multi-hop non è facile da imparare direttamente e richiede una supervisione diretta, inoltre fare bene su WikiHop non significa necessariamente che il modello stia effettivamente imparando a fare ragionamenti multi-hop.", "L'articolo si propone di indagare il ben noto problema dell'apprendimento delle memory network e più precisamente la difficoltà della supervisione dell'apprendimento con attention di tali modelli.", "Questo articolo sostiene che le memory network non riescono ad apprendere un ragionevole ragionamento multi-hop."]} +{"source": "Generative Adversarial Nets (GANs) and Variational Auto-Encoders (VAEs) provide impressive image generations from Gaussian white noise, but the underlying mathematics are not well understood. We compute deep convolutional network generators by inverting a fixed embedding operator. Therefore, they do not require to be optimized with a discriminator or an encoder. The embedding is Lipschitz continuous to deformations so that generators transform linear interpolations between input white noise vectors into deformations between output images. This embedding is computed with a wavelet Scattering transform. Numerical experiments demonstrate that the resulting Scattering generators have similar properties as GANs or VAEs, without learning a discriminative network or an encoder.", "target": ["Introduciamo reti generative che non richiedono di essere apprese con un discriminatore o un encoder; si ottengono invertendo uno speciale operatore di embedding definito da una trasformata wavelet Scattering.", "Introduce le trasformate di scattering come modelli generativi di immagini nel contesto delle Generative Adversarial Network e suggerisce perché potrebbero essere viste come trasformate di Gaussianizzazione con loss di informazione controllata e invertibilità.", "L'articolo propone un modello generativo per le immagini che non richiede l'apprendimento di un discriminatore (come nelle GAN) o di un embedding appreso."]} +{"source": "Recurrent neural networks (RNNs) can model natural language by sequentially ''reading'' input tokens and outputting a distributed representation of each token. Due to the sequential nature of RNNs, inference time is linearly dependent on the input length, and all inputs are read regardless of their importance. Efforts to speed up this inference, known as ''neural speed reading'', either ignore or skim over part of the input. We present Structural-Jump-LSTM: the first neural speed reading model to both skip and jump text during inference. The model consists of a standard LSTM and two agents: one capable of skipping single words when reading, and one capable of exploiting punctuation structure (sub-sentence separators (,:), sentence end symbols (.!?), or end of text markers) to jump ahead after reading a word. A comprehensive experimental evaluation of our model against all five state-of-the-art neural reading models shows that Structural-Jump-LSTM achieves the best overall floating point operations (FLOP) reduction (hence is faster), while keeping the same accuracy or even improving it compared to a vanilla LSTM that reads the whole text.", "target": ["Proponiamo un nuovo modello per neural speed reading che utilizza la struttura intrinseca della punteggiatura di un testo per definire un comportamento efficace di jumping e skipping.", "L'articolo propone un modello Structural-Jump-LSTM per accelerare il machine reading con due agenti invece di uno", "Propone un nuovo modello per neural speed reading in cui il nuovo reader ha la capacità di saltare una parola o una sequenza di parole.", "L'articolo propone un metodo di lettura veloce usando azioni di skip e di jump, mostrando che il metodo proposto è accurato come le LSTM ma usa molto meno computazione."]} +{"source": "One of the key challenges of session-based recommender systems is to enhance users’ purchase intentions. In this paper, we formulate the sequential interactions between user sessions and a recommender agent as a Markov Decision Process (MDP). In practice, the purchase reward is delayed and sparse, and may be buried by clicks, making it an impoverished signal for policy learning. Inspired by the prediction error minimization (PEM) and embodied cognition, we propose a simple architecture to augment reward, namely Imagination Reconstruction Network (IRN). Specifically, IRN enables the agent to explore its environment and learn predictive representations via three key components. The imagination core generates predicted trajectories, i.e., imagined items that users may purchase. The trajectory manager controls the granularity of imagined trajectories using the planning strategies, which balances the long-term rewards and short-term rewards. To optimize the action policy, the imagination-augmented executor minimizes the intrinsic imagination error of simulated trajectories by self-supervised reconstruction, while maximizing the extrinsic reward using model-free algorithms. Empirically, IRN promotes quicker adaptation to user interest, and shows improved robustness to the cold-start scenario and ultimately higher purchase performance compared to several baselines. Somewhat surprisingly, IRN using only the purchase reward achieves excellent next-click prediction performance, demonstrating that the agent can \"guess what you like\" via internal planning.", "target": ["Proponiamo l'architettura IRN per aumentare la reward di acquisto sparsa e ritardata per la raccomandazione basata sulla sessione.", "L'articolo propone di migliorare le prestazioni dei sistemi di raccomandazione attraverso reinforcement learning utilizzando una Imagination Reconstruction Network.", "L'articolo presenta un approccio di raccomandazione basato sulla sessione, concentrandosi sugli acquisti degli utenti invece che sui clic."]} +{"source": "The question why deep learning algorithms generalize so well has attracted increasing research interest. However, most of the well-established approaches, such as hypothesis capacity, stability or sparseness, have not provided complete explanations (Zhang et al., 2016; Kawaguchi et al., 2017). In this work, we focus on the robustness approach (Xu & Mannor, 2012), i.e., if the error of a hypothesis will not change much due to perturbations of its training examples, then it will also generalize well. As most deep learning algorithms are stochastic (e.g., Stochastic Gradient Descent, Dropout, and Bayes-by-backprop), we revisit the robustness arguments of Xu & Mannor, and introduce a new approach – ensemble robustness – that concerns the robustness of a population of hypotheses. Through the lens of ensemble robustness, we reveal that a stochastic learning algorithm can generalize well as long as its sensitiveness to adversarial perturbations is bounded in average over training examples. Moreover, an algorithm may be sensitive to some adversarial examples (Goodfellow et al., 2015) but still generalize well. To support our claims, we provide extensive simulations for different deep learning algorithms and different network architectures exhibiting a strong correlation between ensemble robustness and the ability to generalize.", "target": ["Spiegare la generalizzazione degli algoritmi di deep learning stocastico, teoricamente ed empiricamente, attraverso la robustezza dell'ensemble", "Questo articolo presenta un adattamento della robustezza algoritmica di Xu&Mannor'12 e presenta bound di apprendimento e un esperimento che mostra la correlazione tra la robustezza empirica dell'ensemble e l'errore di generalizzazione.", "Propone uno studio della capacità di generalizzazione degli algoritmi di deep learning utilizzando un'estensione della nozione di stabilità chiamata robustezza dell'ensemble e fornisce bound all'errore di generalizzazione di un algoritmo randomizzato in termini del parametro di stabilità e fornisce uno studio empirico che cerca di collegare la teoria con la pratica.", "L'articolo ha studiato la capacità di generalizzazione degli algoritmi di apprendimento dal punto di vista della robustezza in un contesto di deep learning"]} +{"source": "Deep autoregressive models have shown state-of-the-art performance in density estimation for natural images on large-scale datasets such as ImageNet. However, such models require many thousands of gradient-based weight updates and unique image examples for training. Ideally, the models would rapidly learn visual concepts from only a handful of examples, similar to the manner in which humans learns across many vision tasks. In this paper, we show how 1) neural attention and 2) meta learning techniques can be used in combination with autoregressive models to enable effective few-shot density estimation. Our proposed modifications to PixelCNN result in state-of-the art few-shot density estimation on the Omniglot dataset. Furthermore, we visualize the learned attention policy and find that it learns intuitive algorithms for simple tasks such as image mirroring on ImageNet and handwriting on Omniglot without supervision. Finally, we extend the model to natural images and demonstrate few-shot image generation on the Stanford Online Products dataset.", "target": ["Few-shot learning PixelCNN", "L'articolo propone di utilizzare la stima della densità quando la disponibilità di dati di training è bassa, utilizzando un modello di meta-learning.", "Questo articolo considera il problema della stima della densità in un setting one/few-shot, usando tecniche di metalearning che sono state applicate all'apprendimento supervisionato one/few-shot", "L'articolo si concentra sul few shot learning con la stima della densità autoregressiva e migliora PixelCNN con l'attention neurale e le tecniche di meta learning."]} +{"source": "Neural networks exhibit good generalization behavior in the over-parameterized regime, where the number of network parameters exceeds the number of observations. Nonetheless, current generalization bounds for neural networks fail to explain this phenomenon. In an attempt to bridge this gap, we study the problem of learning a two-layer over-parameterized neural network, when the data is generated by a linearly separable function. In the case where the network has Leaky ReLU activations, we provide both optimization and generalization guarantees for over-parameterized networks. Specifically, we prove convergence rates of SGD to a global minimum and provide generalization guarantees for this global minimum that are independent of the network size. Therefore, our result clearly shows that the use of SGD for optimization both finds a global minimum, and avoids overfitting despite the high capacity of the model. This is the first theoretical demonstration that SGD can avoid overfitting, when learning over-specified neural network classifiers.", "target": ["Mostriamo che SGD apprende reti neurali a due layer iperparametrizzate con attivazioni Leaky ReLU che generalizzano in modo dimostrabile su dati linearmente separabili.", "L'articolo studia i modelli iperparametrizzati in grado di apprendere soluzioni che ben generalizzano utilizzando una rete a 1 hidden layer con un layer di output fisso.", "Questo articolo mostra che, su dati linearmente separabili, SGD su una rete iperparametrizzata può ancora appoggiare un classificatore che generalizza in modo dimostrabile."]} +{"source": "A central challenge in reinforcement learning is discovering effective policies for tasks where rewards are sparsely distributed. We postulate that in the absence of useful reward signals, an effective exploration strategy should seek out {\\it decision states}. These states lie at critical junctions in the state space from where the agent can transition to new, potentially unexplored regions. We propose to learn about decision states from prior experience. By training a goal-conditioned model with an information bottleneck, we can identify decision states by examining where the model accesses the goal state through the bottleneck. We find that this simple mechanism effectively identifies decision states, even in partially observed settings. In effect, the model learns the sensory cues that correlate with potential subgoals. In new environments, this model can then identify novel subgoals for further exploration, guiding the agent through a sequence of potential decision states and through new regions of the state space.", "target": ["Il training degli agenti con goal-policy information bottleneck promuove il transfer e produce un potente bonus di esplorazione", "Propone la regolarizzazione delle loss di RL standard con l'informazione mutua condizionata negativa per la ricerca della policy in un ambiente di RL multi-goal.", "Questo articolo propone il concetto di stato di decisione e propone una regolarizzazione di divergenza KL per apprendere la struttura dei task e usare questa informazione per incoraggiare la policy a visitare gli stati di decisione.", "L'articolo propone un metodo per regolarizzare le policy goal-conditioned con un termine di informazione mutua."]} +{"source": "Many applications in machine learning require optimizing a function whose true gradient is unknown, but where surrogate gradient information (directions that may be correlated with, but not necessarily identical to, the true gradient) is available instead. This arises when an approximate gradient is easier to compute than the full gradient (e.g. in meta-learning or unrolled optimization), or when a true gradient is intractable and is replaced with a surrogate (e.g. in certain reinforcement learning applications or training networks with discrete variables). We propose Guided Evolutionary Strategies, a method for optimally using surrogate gradient directions along with random search. We define a search distribution for evolutionary strategies that is elongated along a subspace spanned by the surrogate gradients. This allows us to estimate a descent direction which can then be passed to a first-order optimizer. We analytically and numerically characterize the tradeoffs that result from tuning how strongly the search distribution is stretched along the guiding subspace, and use this to derive a setting of the hyperparameters that works well across problems. Finally, we apply our method to example problems including truncated unrolled optimization and training neural networks with discrete variables, demonstrating improvement over both standard evolutionary strategies and first-order methods (that directly follow the surrogate gradient). We provide a demo of Guided ES at: redacted URL", "target": ["Proponiamo un metodo di ottimizzazione per quando sono disponibili solo gradienti biased - definiamo un nuovo stimatore di gradiente per questo scenario, deriviamo il bias e la varianza di questo stimatore e lo applichiamo a problemi di esempio.", "Gli autori propongono un approccio che combina la ricerca casuale con l'informazione del gradiente surrogato e fanno una discussione sul trade-off variance-bias così come sull'ottimizzazione degli iperparametri.", "L'articolo propone un metodo per migliorare la ricerca casuale costruendo un sottospazio dei precedenti k gradienti surrogati.", "Questo articolo cerca di accelerare l'evoluzione del tipo OpenAI introducendo una distribuzione non isotropica con una matrice di covarianza nella forma I + UU^t e informazioni esterne come un gradiente surrogato per determinare U"]} +{"source": "Point clouds are an important type of geometric data and have widespread use in computer graphics and vision. However, learning representations for point clouds is particularly challenging due to their nature as being an unordered collection of points irregularly distributed in 3D space. Graph convolution, a generalization of the convolution operation for data defined over graphs, has been recently shown to be very successful at extracting localized features from point clouds in supervised or semi-supervised tasks such as classification or segmentation. This paper studies the unsupervised problem of a generative model exploiting graph convolution. We focus on the generator of a GAN and define methods for graph convolution when the graph is not known in advance as it is the very output of the generator. The proposed architecture learns to generate localized features that approximate graph embeddings of the output geometry. We also study the problem of defining an upsampling layer in the graph-convolutional generator, such that it learns to exploit a self-similarity prior on the data distribution to sample more effectively.", "target": ["Una GAN che usa operazioni di convoluzione dei grafi con grafi calcolati dinamicamente da feature nascoste", "L'articolo propone una versione delle GAN specificamente progettata per la generazione di nuvole di punti con il contributo centrale del lavoro costituito dall'operazione di upsampling.", "Questo articolo propone delle GAN grafo-convoluzionali per nuvole di punti 3D irregolari che imparano il dominio e le feature allo stesso tempo."]} +{"source": "Memorization in over-parameterized neural networks can severely hurt generalization in the presence of mislabeled examples. However, mislabeled examples are to hard avoid in extremely large datasets. We address this problem using the implicit regularization effect of stochastic gradient descent with large learning rates, which we find to be able to separate clean and mislabeled examples with remarkable success using loss statistics. We leverage this to identify and on-the-fly discard mislabeled examples using a threshold on their losses. This leads to On-the-fly Data Denoising (ODD), a simple yet effective algorithm that is robust to mislabeled examples, while introducing almost zero computational overhead. Empirical results demonstrate the effectiveness of ODD on several datasets containing artificial and real-world mislabeled examples.", "target": ["Introduciamo un algoritmo veloce e facile da implementare che è robusto al rumore del dataset.", "Il paper mira a rimuovere i potenziali esempi con rumore sulle label scartando quelli con grandi loss nella procedura di training."]} +{"source": "Binarized Neural Networks (BNNs) have recently attracted significant interest due to their computational efficiency. Concurrently, it has been shown that neural networks may be overly sensitive to ``attacks\" -- tiny adversarial changes in the input -- which may be detrimental to their use in safety-critical domains. Designing attack algorithms that effectively fool trained models is a key step towards learning robust neural networks. The discrete, non-differentiable nature of BNNs, which distinguishes them from their full-precision counterparts, poses a challenge to gradient-based attacks. In this work, we study the problem of attacking a BNN through the lens of combinatorial and integer optimization. We propose a Mixed Integer Linear Programming (MILP) formulation of the problem. While exact and flexible, the MILP quickly becomes intractable as the network and perturbation space grow. To address this issue, we propose IProp, a decomposition-based algorithm that solves a sequence of much smaller MILP problems. Experimentally, we evaluate both proposed methods against the standard gradient-based attack (PGD) on MNIST and Fashion-MNIST, and show that IProp performs favorably compared to PGD, while scaling beyond the limits of the MILP.", "target": ["Gli attacchi basati sul gradiente sulle reti neurali binarizzate non sono efficaci a causa della non differenziabilità di tali reti; il nostro algoritmo IPROP risolve questo problema utilizzando l'ottimizzazione intera", "Propone un nuovo algoritmo in stile propagazione degli obiettivi per generare forti adversarial attack sulle reti neurali binarizzate.", "Questo articolo ha proposto un nuovo algoritmo di attacco basato su MILP su reti neurali binarie.", "Questo articolo presenta un algoritmo per trovare adversarial attack alle reti neurali binarie che trova iterativamente le rappresentazioni desiderate layer per layer dall'alto all'input ed è più efficiente che risolvere il risolutore completo di programmazione lineare intera mista."]} +{"source": "Highly regularized LSTMs achieve impressive results on several benchmark datasets in language modeling. We propose a new regularization method based on decoding the last token in the context using the predicted distribution of the next token. This biases the model towards retaining more contextual information, in turn improving its ability to predict the next token. With negligible overhead in the number of parameters and training time, our Past Decode Regularization (PDR) method achieves a word level perplexity of 55.6 on the Penn Treebank and 63.5 on the WikiText-2 datasets using a single softmax. We also show gains by using PDR in combination with a mixture-of-softmaxes, achieving a word level perplexity of 53.8 and 60.5 on these datasets. In addition, our method achieves 1.169 bits-per-character on the Penn Treebank Character dataset for character level language modeling. These results constitute a new state-of-the-art in their respective settings.", "target": ["La decodifica dell'ultimo token nel contesto usando la distribuzione del token successivo previsto agisce come un regolarizzatore e migliora il language modeling.", "Gli autori introducono l'idea del past decoding allo scopo di regolarizzare la perplexity su Penn Treebank", "Propone un termine di loss aggiuntivo da usare quando si addestra un LM basato su LSTM e mostra che aggiungendo questo termine di loss si può raggiungere la perplexity SOTA su una serie di benchmark LM.", "Suggerisce una nuova tecnica di regolarizzazione che può essere aggiunta in cima a quelle usate in AWD-LSTM di Merity et al. (2017) con poco overhead."]} +{"source": "The assumption that data samples are independently identically distributed is the backbone of many learning algorithms. Nevertheless, datasets often exhibit rich structures in practice, and we argue that there exist some unknown orders within the data instances. Aiming to find such orders, we introduce a novel Generative Markov Network (GMN) which we use to extract the order of data instances automatically. Specifically, we assume that the instances are sampled from a Markov chain. Our goal is to learn the transitional operator of the chain as well as the generation order by maximizing the generation probability under all possible data permutations. One of our key ideas is to use neural networks as a soft lookup table for approximating the possibly huge, but discrete transition matrix. This strategy allows us to amortize the space complexity with a single model and make the transitional operator generalizable to unseen instances. To ensure the learned Markov chain is ergodic, we propose a greedy batch-wise permutation scheme that allows fast training. Empirically, we evaluate the learned Markov chain by showing that GMNs are able to discover orders among data instances and also perform comparably well to state-of-the-art methods on the one-shot recognition benchmark task.", "target": ["Propone di osservare gli ordini impliciti nei dataset nell'ottica di un modello generativo.", "Gli autori affrontano il problema dell'ordinamento implicito in un dataset e la sfida di recuperarlo e propongono di imparare un modello senza distanza metrica che assume una catena di Markov come meccanismo generativo dei dati", "L'articolo propone Generative Markov Network - un approccio basato sul deep learning per modellare le sequenze e scoprire l'ordine nei dataset.", "Propone di imparare l'ordine di un sample di dati non ordinato imparando una catena di Markov."]} +{"source": "We present a Neural Program Search, an algorithm to generate programs from natural language description and a small number of input / output examples. The algorithm combines methods from Deep Learning and Program Synthesis fields by designing rich domain-specific language (DSL) and defining efficient search algorithm guided by a Seq2Tree model on it. To evaluate the quality of the approach we also present a semi-synthetic dataset of descriptions with test examples and corresponding programs. We show that our algorithm significantly outperforms sequence-to-sequence model with attention baseline.", "target": ["Program synthesis dalla descrizione in linguaggio naturale e dagli esempi di input/output tramite la ricerca Tree-Beam sul modello Seq2Tree", "Presenta un modello seq2Tree per tradurre una dichiarazione di un problema in linguaggio naturale al corrispondente programma funzionale in DSL, che ha mostrato un miglioramento rispetto all'approccio di base seq2seq.", "Questo articolo affronta il problema di fare la sintesi di un programma quando viene data una descrizione del problema e un piccolo numero di esempi di input-output.", "L'articolo introduce una tecnica per la sintesi di programmi che coinvolge una grammatica ristretta di problemi che è esplorata tramite beam search usando una rete encoder-decoder con attention."]} +{"source": "Generative adversarial training can be generally understood as minimizing certain moment matching loss defined by a set of discriminator functions, typically neural networks. The discriminator set should be large enough to be able to uniquely identify the true distribution (discriminative), and also be small enough to go beyond memorizing samples (generalizable). In this paper, we show that a discriminator set is guaranteed to be discriminative whenever its linear span is dense in the set of bounded continuous functions. This is a very mild condition satisfied even by neural networks with a single neuron. Further, we develop generalization bounds between the learned distribution and true distribution under different evaluation metrics. When evaluated with neural distance, our bounds show that generalization is guaranteed as long as the discriminator set is small enough, regardless of the size of the generator or hypothesis set. When evaluated with KL divergence, our bound provides an explanation on the counter-intuitive behaviors of testing likelihood in GAN training. Our analysis sheds lights on understanding the practical performance of GANs.", "target": ["Questo articolo studia le proprietà di discriminazione e generalizzazione delle GAN quando l'insieme discriminatore è una classe di funzioni ristretta come le reti neurali.", "Bilancia le capacità delle classi di generatori e discriminatori nelle GAN garantendo che le IPM indotte siano metriche e non pseudo metriche", "Questo articolo fornisce un'analisi matematica del ruolo della dimensione dell'insieme adversarial/discriminatore nelle GAN"]} +{"source": "Normalization layers are a staple in state-of-the-art deep neural network architectures. They are widely believed to stabilize training, enable higher learning rate, accelerate convergence and improve generalization, though the reason for their effectiveness is still an active research topic. In this work, we challenge the commonly-held beliefs by showing that none of the perceived benefits is unique to normalization. Specifically, we propose fixed-update initialization (Fixup), an initialization motivated by solving the exploding and vanishing gradient problem at the beginning of training via properly rescaling a standard initialization. We find training residual networks with Fixup to be as stable as training with normalization -- even for networks with 10,000 layers. Furthermore, with proper regularization, Fixup enables residual networks without normalization to achieve state-of-the-art performance in image classification and machine translation.", "target": ["Tutto ciò di cui avete bisogno per deep residual network è una buona inizializzazione; i layer di normalizzazione non sono necessari.", "Viene presentato un metodo per l'inizializzazione e la normalizzazione delle deep residual network. Questo si basa sulle osservazioni dell'esplosione forward e backward in tali reti. La performance del metodo è alla pari con i migliori risultati ottenuti da altre reti con una normalizzazione più esplicita.", "Gli autori propongono un nuovo modo di inizializzare le residual network, che è motivato dalla necessità di evitare gli exploding/vanishing gradient.", "Propone un nuovo metodo di inizializzazione usato per addestrare ResNet molto profonde senza usare batch-norm."]} +{"source": "Designing a metric manually for unsupervised sequence generation tasks, such as text generation, is essentially difficult. In a such situation, learning a metric of a sequence from data is one possible solution. The previous study, SeqGAN, proposed the framework for unsupervised sequence generation, in which a metric is learned from data, and a generator is optimized with regard to the learned metric with policy gradient, inspired by generative adversarial nets (GANs) and reinforcement learning. In this paper, we make two proposals to learn better metric than SeqGAN's: partial reward function and expert-based reward function training. The partial reward function is a reward function for a partial sequence of a certain length. SeqGAN employs a reward function for completed sequence only. By combining long-scale and short-scale partial reward functions, we expect a learned metric to be able to evaluate a partial correctness as well as a coherence of a sequence, as a whole. In expert-based reward function training, a reward function is trained to discriminate between an expert (or true) sequence and a fake sequence that is produced by editing an expert sequence. Expert-based reward function training is not a kind of GAN frameworks. This makes the optimization of the generator easier. We examine the effect of the partial reward function and expert-based reward function training on synthetic data and real text data, and show improvements over SeqGAN and the model trained with MLE. Specifically, whereas SeqGAN gains 0.42 improvement of NLL over MLE on synthetic data, our best model gains 3.02 improvement, and whereas SeqGAN gains 0.029 improvement of BLEU over MLE, our best model gains 0.250 improvement.", "target": ["Questo articolo mira a imparare una metrica migliore per l'apprendimento unsupervised, come la generazione di testo, e mostra un miglioramento significativo rispetto a SeqGAN.", "Descrive un approccio alla generazione di sequenze temporali attraverso l'apprendimento di valori stato-azione, dove lo stato è la sequenza generata finora, e l'azione è la scelta del valore successivo.", "Questo articolo considera il problema di migliorare la generazione di sequenze imparando metriche migliori, in particolare il problema del bias di esposizione"]} +{"source": "One of the most successful techniques in generative models has been decomposing a complicated generation task into a series of simpler generation tasks. For example, generating an image at a low resolution and then learning to refine that into a high resolution image often improves results substantially. Here we explore a novel strategy for decomposing generation for complicated objects in which we first generate latent variables which describe a subset of the observed variables, and then map from these latent variables to the observed space. We show that this allows us to achieve decoupled training of complicated generative models and present both theoretical and experimental results supporting the benefit of such an approach.", "target": ["Decompone il task di apprendere un modello generativo nell'apprendimento di fattori latenti disgiunti per sottoinsiemi di dati per poi apprendere la distribuzione congiunta su quei fattori latenti.", "Fattori Localmente Disentangled per un modello generativo gerarchico a variabili latenti, che può essere visto come una variante gerarchica di Adversarially Learned Inference", "L'articolo indaga il potenziale dei modelli gerarchici a variabili latenti per generare immagini e sequenze di immagini e propone di addestrare diversi modelli ALI impilati uno sull'altro per creare una rappresentazione gerarchica dei dati.", "L'articolo mira ad apprendere le gerarchie per il training delle GAN direttamente in un programma di ottimizzazione gerarchica invece di essere progettate da un umano"]} +{"source": "Visual grounding of language is an active research field aiming at enriching text-based representations with visual information. In this paper, we propose a new way to leverage visual knowledge for sentence representations. Our approach transfers the structure of a visual representation space to the textual space by using two complementary sources of information: (1) the cluster information: the implicit knowledge that two sentences associated with the same visual content describe the same underlying reality and (2) the perceptual information contained within the structure of the visual space. We use a joint approach to encourage beneficial interactions during training between textual, perceptual, and cluster information. We demonstrate the quality of the learned representations on semantic relatedness, classification, and cross-modal retrieval tasks.", "target": ["Proponiamo un modello congiunto per incorporare la conoscenza visiva nelle rappresentazioni delle frasi", "L'articolo propone un metodo per utilizzare i video abbinati a caption per migliorare i sentence embedding", "Questo paper propone un modello per l'apprendimento di rappresentazioni di frasi che sono fondate e basate su dati video associati.", "Propone un metodo per migliorare i sentence embedding basati sul testo attraverso un framework multimodale congiunto."]} \ No newline at end of file