Datasets:

ahmedBargady

/

arxiv-abstracts-with-categories

like 0

In-context learning provides a new perspective for multi-task modeling for vision and NLP. Under this setting, the model can perceive tasks from prompts and accomplish them without any extra task-specific head predictions or model fine-tuning. However, Skeleton sequence modeling via in-context learning remains unexplored. Directly applying existing in-context models from other areas onto skeleton sequences fails due to the inter-frame and cross-task pose similarity that makes it outstandingly hard to perceive the task correctly from a subtle context. To address this challenge, we propose Skeleton-in-Context (SiC), an effective framework for in-context skeleton sequence modeling. Our SiC is able to handle multiple skeleton-based tasks simultaneously after a single training process and accomplish each task from context according to the given prompt. It can further generalize to new, unseen tasks according to customized prompts. To facilitate context perception, we additionally propose a task-unified prompt, which adaptively learns tasks of different natures, such as partial joint-level generation, sequence-level prediction, or 2D-to-3D motion prediction. We conduct extensive experiments to evaluate the effectiveness of our SiC on multiple tasks, including motion prediction, pose estimation, joint completion, and future pose estimation. We also evaluate its generalization capability on unseen tasks such as motion-in-between. These experiments show that our model achieves state-of-the-art multi-task performance and even outperforms single-task methods on certain tasks.

cs

In this paper, we propose a general numerical framework to derive structure-preserving reduced order models for thermodynamically consistent PDEs. Our numerical framework has two primary features: (a) a systematic way to extract reduced order models for thermodynamically consistent PDE systems while maintaining their inherent thermodynamic principles and (b) a strategic process to devise accurate, efficient, and structure-preserving numerical algorithms to solve the forehead reduced-order models. The platform's generality extends to various PDE systems governed by embedded thermodynamic laws. The proposed numerical platform is unique from several perspectives. First, it utilizes the generalized Onsager principle to transform the thermodynamically consistent PDE system into an equivalent one, where the transformed system's free energy adopts a quadratic form of the state variables. This transformation is named energy quadratization (EQ). Through EQ, we gain a novel perspective on deriving reduced order models. The reduced order models derived through our method continue to uphold the energy dissipation law. Secondly, our proposed numerical approach automatically provides numerical algorithms to discretize the reduced order models. The proposed algorithms are always linear, easy to implement and solve, and uniquely solvable. Furthermore, these algorithms inherently ensure the thermodynamic laws. In essence, our platform offers a distinctive approach to derive structure-preserving reduced-order models for a wide range of PDE systems abiding by thermodynamic principles.

math

With increasing scale in model and dataset size, the training of deep neural networks becomes a massive computational burden. One approach to speed up the training process is Selective Backprop. For this approach, we perform a forward pass to obtain a loss value for each data point in a minibatch. The backward pass is then restricted to a subset of that minibatch, prioritizing high-loss examples. We build on this approach, but seek to improve the subset selection mechanism by choosing the (weighted) subset which best matches the mean gradient over the entire minibatch. We use the gradients w.r.t. the model's last layer as a cheap proxy, resulting in virtually no overhead in addition to the forward pass. At the same time, for our experiments we add a simple random selection baseline which has been absent from prior work. Surprisingly, we find that both the loss-based as well as the gradient-matching strategy fail to consistently outperform the random baseline.

math

Meta-learning enables rapid generalization to new tasks by learning meta-knowledge from a variety of tasks. It is intuitively assumed that the more tasks a model learns in one training batch, the richer knowledge it acquires, leading to better generalization performance. However, contrary to this intuition, our experiments reveal an unexpected result: adding more tasks within a single batch actually degrades the generalization performance. To explain this unexpected phenomenon, we conduct a Structural Causal Model (SCM) for causal analysis. Our investigation uncovers the presence of spurious correlations between task-specific causal factors and labels in meta-learning. Furthermore, the confounding factors differ across different batches. We refer to these confounding factors as ``Task Confounders". Based on this insight, we propose a plug-and-play Meta-learning Causal Representation Learner (MetaCRL) to eliminate task confounders. It encodes decoupled causal factors from multiple tasks and utilizes an invariant-based bi-level optimization mechanism to ensure their causality for meta-learning. Extensive experiments on various benchmark datasets demonstrate that our work achieves state-of-the-art (SOTA) performance.

cs

This paper investigates the problem of designing control policies that satisfy high-level specifications described by signal temporal logic (STL) in unknown, stochastic environments. While many existing works concentrate on optimizing the spatial robustness of a system, our work takes a step further by also considering temporal robustness as a critical metric to quantify the tolerance of time uncertainty in STL. To this end, we formulate two relevant control objectives to enhance the temporal robustness of the synthesized policies. The first objective is to maximize the probability of being temporally robust for a given threshold. The second objective is to maximize the worst-case spatial robustness value within a bounded time shift. We use reinforcement learning to solve both control synthesis problems for unknown systems. Specifically, we approximate both control objectives in a way that enables us to apply the standard Q-learning algorithm. Theoretical bounds in terms of the approximations are also derived. We present case studies to demonstrate the feasibility of our approach.

cs

Large Language Models (LLMs) have demonstrated remarkable capabilities in important tasks such as natural language understanding, language generation, and complex reasoning and have the potential to make a substantial impact on our society. Such capabilities, however, come with the considerable resources they demand, highlighting the strong need to develop effective techniques for addressing their efficiency challenges. In this survey, we provide a systematic and comprehensive review of efficient LLMs research. We organize the literature in a taxonomy consisting of three main categories, covering distinct yet interconnected efficient LLMs topics from model-centric, data-centric, and framework-centric perspective, respectively. We have also created a GitHub repository where we compile the papers featured in this survey at , , and will actively maintain this repository and incorporate new research as it emerges. We hope our survey can serve as a valuable resource to help researchers and practitioners gain a systematic understanding of the research developments in efficient LLMs and inspire them to contribute to this important and exciting field.

cs

Personalizing text-to-image models using a limited set of images for a specific object has been explored in subject-specific image generation. However, existing methods often encounter challenges in aligning with text prompts due to overfitting to the limited training images. In this work, we introduce InstructBooth, a novel method designed to enhance image-text alignment in personalized text-to-image models. Our approach first personalizes text-to-image models with a small number of subject-specific images using a unique identifier. After personalization, we fine-tune personalized text-to-image models using reinforcement learning to maximize a reward that quantifies image-text alignment. Additionally, we propose complementary techniques to increase the synergy between these two processes. Our method demonstrates superior image-text alignment compared to baselines while maintaining personalization ability. In human evaluations, InstructBooth outperforms DreamBooth when considering all comprehensive factors.

cs

Open-world 3D part segmentation is pivotal in diverse applications such as robotics and AR/VR. Traditional supervised methods often grapple with limited 3D data availability and struggle to generalize to unseen object categories. PartSLIP, a recent advancement, has made significant strides in zero- and few-shot 3D part segmentation. This is achieved by harnessing the capabilities of the 2D open-vocabulary detection module, GLIP, and introducing a heuristic method for converting and lifting multi-view 2D bounding box predictions into 3D segmentation masks. In this paper, we introduce PartSLIP++, an enhanced version designed to overcome the limitations of its predecessor. Our approach incorporates two major improvements. First, we utilize a pre-trained 2D segmentation model, SAM, to produce pixel-wise 2D segmentations, yielding more precise and accurate annotations than the 2D bounding boxes used in PartSLIP. Second, PartSLIP++ replaces the heuristic 3D conversion process with an innovative modified Expectation-Maximization algorithm. This algorithm conceptualizes 3D instance segmentation as unobserved latent variables, and then iteratively refines them through an alternating process of 2D-3D matching and optimization with gradient descent. Through extensive evaluations, we show that PartSLIP++ demonstrates better performance over PartSLIP in both low-shot 3D semantic and instance-based object part segmentation tasks. Code released at .

cs

Semantic communications are expected to become the core new paradigms of the sixth generation (6G) wireless networks. Most existing works implicitly utilize channel information for codecs training, which leads to poor communications when channel type or statistical characteristics change. To tackle this issue posed by various channels, a novel channel-transferable semantic communications (CT-SemCom) framework is proposed, which adapts the codecs learned on one type of channel to other types of channels. Furthermore, integrating the proposed framework and the orthogonal frequency division multiplexing systems integrating non-orthogonal multiple access technologies, i.e., OFDM-NOMA systems, a power allocation problem to realize the transfer from additive white Gaussian noise (AWGN) channels to multi-subcarrier Rayleigh fading channels is formulated. We then design a semantics-similar dual transformation (SSDT) algorithm to derive analytical solutions with low complexity. Simulation results show that the proposed CT-SemCom framework with SSDT algorithm significantly outperforms the existing work w.r.t. channel transferability, e.g., the peak signal-to-noise ratio (PSNR) of image transmission improves by 4.2-7.3 dB under different variances of Rayleigh fading channels.

cs

In this study, we carry out a non-perturbative approach to a quantum Otto engine, employing an Unruh-DeWitt particle detector to extract work from a quantum Klein-Gordon field in an arbitrary globally hyperbolic curved spacetime. We broaden the scope by considering the field in any quasi-free state, which includes vacuum, thermal, and squeezed states. A key aspect of our method is the instantaneous interaction between the detector and the field, which enables a thorough non-perturbative analysis. We demonstrate that the detector can successfully extract positive work from the quantum Otto cycle, even when two isochoric processes occur instantaneously, provided the detector in the second isochoric process receives a signal from the first interaction. This signaling allows the detector to release heat into the field, thereby the thermodynamic cycle is completed. As a demonstration, we consider a detector at rest in flat spacetime and compute the work extracted from the Minkowski vacuum state.

quant-ph

Opinion dynamics is crucial for unraveling the complexities of human interaction in the information age. How to speed up consensus without disturbing the fate of the system is key for opinion dynamics. We propose a voter model on adaptive networks, which resembles the coevolutionary process between opinions and social relationships. We prove the existence of a one-dimensional stable manifold for the system, which facilitates us to study both the fate of the system and the consensus time it takes. Surprisingly, we find the adjustment of social relations speeds up consensus but does not affect the fate of the system. For echo-chamber-like networks which consist of two homogeneous subnetworks connected by few sparse links, a small probability of adaptive edge dynamics is sufficient to accelerate consensus formation, which is counterintuitive. If the network structure makes consensus much slower than that of the regular networks, minor random rewiring makes a discontinuous drop in consensus time. Our work opens up an avenue for speeding up consensus without disturbing the fate of the system. It can be insightful for crowd control.

math

We prove a weak iterated invariance principle for a large class of non-uniformly expanding random dynamical systems. In addition, we give a quenched homogenization result for fast-slow systems in the case when the fast component corresponds to a uniformly expanding random system. Our techniques rely on the appropriate martingale decomposition.

math

We construct a compact Hausdorff space $K$ such that the space $P(K)$ of Radon probabiblity measures on $K$ considered with the weak$^*$ topology (induced from the space of continuous functions $C(K)$) is countably tight which is a generalization of sequentiality (i.e., if a measure $\mu$ is in the closure of a set $M$, there is a countable $M'\subseteq M$ such that $\mu$ is in the closure of $M'$) but $K$ carries a Radon probability measure which has uncountable Maharam type (i.e., $L_1(\mu)$ is nonseparable). The construction uses (necessarily) an additional set-theoretic assumption (the $\diamondsuit$ principle) as it was already known, by a result of Fremlin, that it is consistent that such spaces do not exist. This should be compared with the result of Plebanek and Sobota who showed that countable tightness of $P(K\times K)$ implies that all Radon measures on $K$ have countable type. So, our example shows that the tightness of $P(K\times K)$ and of $P(K)\times P(K)$ can be different as well as $P(K)$ may have Corson property (C) while $P(K\times K)$ fails to have it answering a question of Pol. Our construction is also a relevant example in the general context of injective tensor products of Banach spaces complementing recent results of Avilés, Martínez-Cervantes, Rodríguez and Rueda Zoca.

math

In the context of sketching for compressive mixture modeling, we revisit existing proofs of the Restricted Isometry Property of sketching operators with respect to certain mixtures models. After examining the shortcomings of existing guarantees, we propose an alternative analysis that circumvents the need to assume importance sampling when drawing random Fourier features to build random sketching operators. Our analysis is based on new deterministic bounds on the restricted isometry constant that depend solely on the set of frequencies used to define the sketching operator; then we leverage these bounds to establish concentration inequalities for random sketching operators that lead to the desired RIP guarantees. Our analysis also opens the door to theoretical guarantees for structured sketching with frequencies associated to fast random linear operators.

cs

Breast ultrasound (BUS) is a critical diagnostic tool in the field of breast imaging, aiding in the early detection and characterization of breast abnormalities. Interpreting breast ultrasound images commonly involves creating comprehensive medical reports, containing vital information to promptly assess the patient's condition. However, the ultrasound imaging system necessitates capturing multiple images of various parts to compile a single report, presenting a time-consuming challenge. To address this problem, we propose the integration of multiple image analysis tools through a LangChain using Large Language Models (LLM), into the breast reporting process. Through a combination of designated tools and text generation through LangChain, our method can accurately extract relevant features from ultrasound images, interpret them in a clinical context, and produce comprehensive and standardized reports. This approach not only reduces the burden on radiologists and healthcare professionals but also enhances the consistency and quality of reports. The extensive experiments shows that each tools involved in the proposed method can offer qualitatively and quantitatively significant results. Furthermore, clinical evaluation on the generated reports demonstrates that the proposed method can make report in clinically meaningful way.

cs

In this paper we argue that topic plays a fundamental role in conversations, and that the concept is needed in addition to that of genre to define interactions. In particular, the concepts of genre and topic need to be separated and orthogonally defined. This would enable modular, reliable and controllable flexible-domain dialogue systems.

cs

Recently, some application (L7) processing has been moved to the network stack (including proxies) as a way to provide a common and application-agnostic interface for security policies, simplify service management, etc. This paper looks at whether L7 network functionality can be offloaded to SmartNICs to improve performance and reduce overheads. We investigate this question by examining how to offload to FPGA NICs the task of L7 dispatch: determining which application process must handle a particular application request. We find that existing protocols, e.g., gRPC on TCP or QUIC, are a hindrance for offloading because the data required to make L7 dispatch decisions can be spread out over many packets, requiring the NIC to buffer and parse a large number of packets to reconstruct L7 data. This paper presents QingNiao-an approach that co-designs application libraries, network protocols, and hardware-to make L7 dispatch feasible under NIC memory constraints. We prototype QingNiao on a 100Gbps FPGA NIC. Experiments with real-world applications show that QingNiao supports a variety of dispatch rules and achieves 7.5x to 8x throughput compared to state-of-the-art software L7 dispatcher. We have open-sourced QingNiao at [1].

cs

Self-driving laboratories (SDLs) consist of multiple stations that perform material synthesis and characterisation tasks. To minimize station downtime and maximize experimental throughput, it is practical to run experiments in asynchronous parallel, in which multiple experiments are being performed at once in different stages. Asynchronous parallelization of experiments, however, introduces delayed feedback (i.e. "pending experiments"), which is known to reduce Bayesian optimiser performance. Here, we build a simulator for a multi-stage SDL and compare optimisation strategies for dealing with delayed feedback and asynchronous parallelized operation. Using data from a real SDL, we build a ground truth Bayesian optimisation simulator from 177 previously run experiments for maximizing the conductivity of functional coatings. We then compare search strategies such as expected improvement, noisy expected improvement, 4-mode exploration and random sampling. We evaluate their performance in terms of amount of delay and problem dimensionality. Our simulation results showcase the trade-off between the asynchronous parallel operation and delayed feedback.

cs

Despite the promise of Neural Posterior Estimation (NPE) methods in astronomy, the adaptation of NPE into the routine inference workflow has been slow. We identify three critical issues: the need for custom featurizer networks tailored to the observed data, the inference inexactness, and the under-specification of physical forward models. To address the first two issues, we introduce a new framework and open-source software nbi (Neural Bayesian Inference), which supports both amortized and sequential NPE. First, nbi provides built-in "featurizer" networks with demonstrated efficacy on sequential data, such as light curve and spectra, thus obviating the need for this customization on the user end. Second, we introduce a modified algorithm SNPE-IS, which facilities asymptotically exact inference by using the surrogate posterior under NPE only as a proposal distribution for importance sampling. These features allow nbi to be applied off-the-shelf to astronomical inference problems involving light curves and spectra. We discuss how nbi may serve as an effective alternative to existing methods such as Nested Sampling. Our package is at .

cs

The silicon vertex detector (SVD) is installed at the heart of the Belle II experiment, taking data at the high-luminosity $B$-Factory SuperKEKB since 2019. The detector has shown a stable and above-99\% hit efficiency, with a large signal-to-noise in all sensors since the beginning of data taking. Cluster position and time resolution have been measured with 2020 and 2022 data and show excellent performance and stability. The effect of radiation damage is visible, but not affecting the performance. As the luminosity increases, higher machine backgrounds are expected and the excellent hit-time information in SVD can be exploited for background rejection. In particular, we have recently developed a novel procedure to select hits by grouping them event-by-event based on their time. This new procedure allows a significant reduction of the fake rate, while preserving the tracking efficiency, and it has therefore replaced the previous cut-based procedure. We have developed a method that uses the SVD hits to estimate the track time (previously unavailable) and the collision time. It has a similar precision to the estimate based on the drift chamber but its execution time is three orders of magnitude smaller, allowing a faster online reconstruction that is crucial in a high luminosity regime. The track time is a powerful information provided to analysis that allows, together with the above-mention grouping selection, to raise the occupancy limit above that expected at nominal luminosity, leaving room for a safety factor. Finally, in June 2022 the data taking of the Belle II experiment was stopped to install a new two-layer DEPFET detector (PXD) and upgrade components of the accelerator. The whole silicon tracker (PXD+SVD) has been extracted from Belle II, the new PXD installed, the detector closed and commissioned. We briefly describe the SVD results of this upgrade.

hep-ex

Beamforming is a powerful tool for physical layer security, as it can be used for steering signals towards legitimate receivers and away from eavesdroppers. An active eavesdropper, however, can interfere with the pilot phase that the transmitter needs to acquire the channel knowledge necessary for beamforming. By doing so, the eavesdropper can make the transmitter form beams towards the eavesdropper rather than towards the legitimate receiver. To mitigate active eavesdroppers, we propose VILLAIN, a novel channel estimator that uses secret pilots. When an eavesdropper interferes with the pilot phase, VILLAIN produces a channel estimate that is orthogonal to the eavesdropper's channel (in the noiseless case). We prove that beamforming based on this channel estimate delivers the highest possible signal power to the legitimate receiver without delivering any signal power to the eavesdropper. Simulations show that VILLAIN mitigates active eavesdroppers also in the noisy case.

cs

Multilinear Principal Component Analysis (MPCA) is a widely utilized method for the dimension reduction of tensor data. However, the integration of MPCA into federated learning remains unexplored in existing research. To tackle this gap, this article proposes a Federated Multilinear Principal Component Analysis (FMPCA) method, which enables multiple users to collaboratively reduce the dimension of their tensor data while keeping each user's data local and confidential. The proposed FMPCA method is guaranteed to have the same performance as traditional MPCA. An application of the proposed FMPCA in industrial prognostics is also demonstrated. Simulated data and a real-world data set are used to validate the performance of the proposed method.

cs

The Green functions were first introduced by Green to compute the character table of GLn(q) in 1955. They were later generalized by Deligne and Lusztig for an arbitrary finite group of Lie type G(q) using l-adic cohomological methods (1976). They proved that these Green functions satisfy an orthogonality relation (we call the first orthogonality relation). Ten years later Kawanaka proved that they satisfy an other orthogonality relation (we call the second orthogonality relation). In this notes, we provide a geometrical understanding of these two orthogonality relations and explain how we can see geometrically that the two orthogonality relations are in fact equivalent.

math

In this paper, we aim to model 3D scene dynamics from multi-view videos. Unlike the majority of existing works which usually focus on the common task of novel view synthesis within the training time period, we propose to simultaneously learn the geometry, appearance, and physical velocity of 3D scenes only from video frames, such that multiple desirable applications can be supported, including future frame extrapolation, unsupervised 3D semantic scene decomposition, and dynamic motion transfer. Our method consists of three major components, 1) the keyframe dynamic radiance field, 2) the interframe velocity field, and 3) a joint keyframe and interframe optimization module which is the core of our framework to effectively train both networks. To validate our method, we further introduce two dynamic 3D datasets: 1) Dynamic Object dataset, and 2) Dynamic Indoor Scene dataset. We conduct extensive experiments on multiple datasets, demonstrating the superior performance of our method over all baselines, particularly in the critical tasks of future frame extrapolation and unsupervised 3D semantic scene decomposition.

cs

With the rise of social media, the spread of fake news has become a significant concern, potentially misleading public perceptions and impacting social stability. Although deep learning methods like CNNs, RNNs, and Transformer-based models like BERT have enhanced fake news detection, they primarily focus on content, overlooking social context during news propagation. Graph-based techniques have incorporated this social context but are limited by the need for large labeled datasets. Addressing these challenges, this paper introduces GAMC, an unsupervised fake news detection technique using the Graph Autoencoder with Masking and Contrastive learning. By leveraging both the context and content of news propagation as self-supervised signals, our method negates the requirement for labeled datasets. We augment the original news propagation graph, encode these with a graph encoder, and employ a graph decoder for reconstruction. A unique composite loss function, including reconstruction error and contrast loss, is designed. The method's contributions are: introducing self-supervised learning to fake news detection, proposing a graph autoencoder integrating two distinct losses, and validating our approach's efficacy through real-world dataset experiments.

cs

In this paper, we obtain a rich family of identities for transposed Poisson $n$-Lie algebras, and then prove the conjecture of Bai, Bai, Guo and Wu in \cite{BBGW} under certain strong condition.

math

Contact tracing is one of the most important tools for preventing the spread of infectious diseases, but as the experience of COVID-19 showed, it is also next-to-impossible to implement when the disease is spreading rapidly. We show how to substantially improve the efficiency of contact tracing by combining standard microeconomic tools that measure heterogeneity in how infectious a sick person is with ideas from machine learning about sequential optimization. Our contributions are twofold. First, we incorporate heterogeneity in individual infectiousness in a multi-armed bandit to establish optimal algorithms. At the heart of this strategy is a focus on learning. In the typical conceptualization of contact tracing, contacts of an infected person are tested to find more infections. Under a learning-first framework, however, contacts of infected persons are tested to ascertain whether the infected person is likely to be a "high infector" and to find additional infections only if it is likely to be highly fruitful. Second, we demonstrate using three administrative contact tracing datasets from India and Pakistan during COVID-19 that this strategy improves efficiency. Using our algorithm, we find 80% of infections with just 40% of contacts while current approaches test twice as many contacts to identify the same number of infections. We further show that a simple strategy that can be easily implemented in the field performs at nearly optimal levels, allowing for, what we call, feasible contact tracing. These results are immediately transferable to contact tracing in any epidemic.

stat

The use of modern software-defined radio (SDR) devices enables the implementation of efficient communication systems in numerous scenarios. Such technology comes especially handy in the context of search and rescue (SAR) systems, enabling the incorporation of additional communication data transmission into the otherwise sub-optimally used SAR bands at 121.5 and 243~MHz. In this work, we propose a novel low-complexity, energy-efficient modulation scheme that allows transmission of additional data within chirped homing signals, while still meeting the standards of international SAR systems such as COSPAS-SARSAT. The proposed method modulates information onto small deviations of the chirp slope with respect to the required unmodulated chirp, which can be easily detected at the receiver side using digital signal processing.

eess

Modern Large Vision-Language Models (LVLMs) enjoy the same vision vocabulary -- CLIP, which can cover most common vision tasks. However, for some special vision task that needs dense and fine-grained vision perception, e.g., document-level OCR or chart understanding, especially in non-English scenarios, the CLIP-style vocabulary may encounter low efficiency in tokenizing the vision knowledge and even suffer out-of-vocabulary problem. Accordingly, we propose Vary, an efficient and effective method to scale up the vision vocabulary of LVLMs. The procedures of Vary are naturally divided into two folds: the generation and integration of a new vision vocabulary. In the first phase, we devise a vocabulary network along with a tiny decoder-only transformer to produce the desired vocabulary via autoregression. In the next, we scale up the vanilla vision vocabulary by merging the new one with the original one (CLIP), enabling the LVLMs can quickly garner new features. Compared to the popular BLIP-2, MiniGPT4, and LLaVA, Vary can maintain its vanilla capabilities while enjoying more excellent fine-grained perception and understanding ability. Specifically, Vary is competent in new document parsing features (OCR or markdown conversion) while achieving 78.2% ANLS in DocVQA and 36.2% in MMVet. Our code will be publicly available on the homepage.

cs

We present a systematic investigation of the quantum games, constructed using a novel repeated game protocol, when played repeatedly ad infinitum. We focus on establishing that such repeated games -- by virtue of inherent quantum-mechanical randomness -- can be mapped to the paradigm of stochastic games. Subsequently, using the setup of two-player--two-action games, we explore the pure reactive strategies belonging to the set of reactive strategies, whose support in the quantum games is no longer countably finite but rather non-denumerably infinite. We find that how two pure strategies fare against each other is crucially dependent on the discount factor (the probability of occurrence of every subsequent round) and how much entangled the quantum states of the players are. We contrast the results obtained with the corresponding results in the classical setup and find fundamental differences between them: e.g, when the underlying game is the prisoner's dilemma, in the quantum game setup, always-defect strategy can be beaten by the tit-for-tat strategy for high enough discount factor.

quant-ph

This paper assesses the potential for large language models (LLMs) to serve as assistive tools in the creative writing process, by means of a single, in-depth case study. In the course of the study, we develop interactive and multi-voice prompting strategies that interleave background descriptions (scene setting, plot elements), instructions that guide composition, samples of text in the target style, and critical discussion of the given samples. We qualitatively evaluate the results from a literary critical perspective, as well as from the standpoint of computational creativity (a sub-field of artificial intelligence). Our findings lend support to the view that the sophistication of the results that can be achieved with an LLM mirrors the sophistication of the prompting.

cs

Despite recent progress in offline learning, these methods are still trained and tested on the same environment. In this paper, we compare the generalization abilities of widely used online and offline learning methods such as online reinforcement learning (RL), offline RL, sequence modeling, and behavioral cloning. Our experiments show that offline learning algorithms perform worse on new environments than online learning ones. We also introduce the first benchmark for evaluating generalization in offline learning, collecting datasets of varying sizes and skill-levels from Procgen (2D video games) and WebShop (e-commerce websites). The datasets contain trajectories for a limited number of game levels or natural language instructions and at test time, the agent has to generalize to new levels or instructions. Our experiments reveal that existing offline learning algorithms struggle to match the performance of online RL on both train and test environments. Behavioral cloning is a strong baseline, outperforming state-of-the-art offline RL and sequence modeling approaches when trained on data from multiple environments and tested on new ones. Finally, we find that increasing the diversity of the data, rather than its size, improves performance on new environments for all offline learning algorithms. Our study demonstrates the limited generalization of current offline learning algorithms highlighting the need for more research in this area.

cs

Multi-goal robot manipulation tasks with sparse rewards are difficult for reinforcement learning (RL) algorithms due to the inefficiency in collecting successful experiences. Recent algorithms such as Hindsight Experience Replay (HER) expedite learning by taking advantage of failed trajectories and replacing the desired goal with one of the achieved states so that any failed trajectory can be utilized as a contribution to learning. However, HER uniformly chooses failed trajectories, without taking into account which ones might be the most valuable for learning. In this paper, we address this problem and propose a novel approach Contact Energy Based Prioritization~(CEBP) to select the samples from the replay buffer based on rich information due to contact, leveraging the touch sensors in the gripper of the robot and object displacement. Our prioritization scheme favors sampling of contact-rich experiences, which are arguably the ones providing the largest amount of information. We evaluate our proposed approach on various sparse reward robotic tasks and compare them with the state-of-the-art methods. We show that our method surpasses or performs on par with those methods on robot manipulation tasks. Finally, we deploy the trained policy from our method to a real Franka robot for a pick-and-place task. We observe that the robot can solve the task successfully. The videos and code are publicly available at:

cs

We study a family of strong fractional integral operators whose kernels have singularity on every coordinate subspace. We prove a two-weight $L^p$-$L^q$-norm inequality by allowing only one of the weights to satisfy $A_p\times A_p$-condition.

math

Cybersecurity in building energy management is crucial for protecting infrastructure, ensuring data integrity, and preventing unauthorized access or manipulation. This paper investigates the energy efficiency and cybersecurity of building energy management systems (BMS) against false data injection (FDI) attacks using proportional-integral (PI) and model predictive control (MPC) methods. Focusing on a commercial building model with five rooms, vulnerability of PI-based BMS and nonlinear MPC-based BMS against FDIs on sensors and actuators is studied. The study aims to assess the effectiveness of these control strategies in maintaining system performance and lifespan, highlighting the potential of MPC in enhancing system resilience against cyber threats. Our case studies demonstrate that even a short term FDIA can cause a 12% reduction in lifetime of a heat-pump under an MPC controller, and cause a near thirty-fold overuse of flow valves under a PI controller.

cs

The SiO emissions are usually used to trace high-velocity outflow shocks in star-forming regions. However, several studies have found low-velocity and widespread SiO emissions not associated with outflows in molecular clouds. We aim to detect and characterize the SiO emissions in massive dense cores (MDCs), and explore the properties of the central sources of SiO emission. We present high-angular-resolution ($\sim$1.5$^{\prime\prime}$) observations of the SiO (5$-$4) line made with the Submillimeter Array towards a sample of 48 MDCs in the Cygnus-X star-forming complex. We studied the SiO emission structures, including their morphologies, kinematics, and energetics, and investigated their relationship with the evolution of the central sources. The SiO (5$-$4) emission is detected in 16 out of 48 MDCs. We identify 14 bipolar and 18 unipolar SiO (5$-$4) outflows associated with 29 dust condensations. Most outflows (24 out of 32) are associated with excess Spitzer 4.5 $\mu$m emissions. We also find diffuse low-velocity ($\Delta{v}$ $\le$ 1.2 km s$^{-1}$) SiO (5$-$4) emission closely surrounding the dust condensations in two MDCs, and suggest that it may originate from decelerated outflow shocks or large-scale shocks from global cloud collapse. We find that the SMA SiO (5$-$4) emission in MDCs is mostly associated with outflows. Probably due to the relatively high excitation of SiO (5$-$4) compared to SiO (2$-$1) and due to the spatial filtering effect, we do not detect large-scale low-velocity SiO (5$-$4) emission, but detect more compact low-velocity emission in close proximity to the dust condensations. We group the sources into different evolutionary stages based on the infrared emission, radio continuum emission, and gas temperature properties of the outflow central sources, and find that the 24 $\mu$m luminosity tends to increase with evolution.

astro-ph

Exact travelling wave solutions to the two-dimensional stochastic Allen-Cahn equation with multiplicative noise are obtained through the hyperbolic tangent (tanh) method. This technique limits the solutions to travelling wave profiles by representing them with a finite tanh power series. This study focuses on how multiplicative noise affects the dynamics of these travelling waves, in particular, occurring of wave propagation failure due to high levels of noise.

math

In order to create machine learning systems that serve a variety of users well, it is vital to not only achieve high average performance but also ensure equitable outcomes across diverse groups. However, most machine learning methods are designed to improve a model's average performance on a chosen end task without consideration for their impact on worst group error. Multitask learning (MTL) is one such widely used technique. In this paper, we seek not only to understand the impact of MTL on worst-group accuracy but also to explore its potential as a tool to address the challenge of group-wise fairness. We primarily consider the common setting of fine-tuning a pre-trained model, where, following recent work (Gururangan et al., 2020; Dery et al., 2023), we multitask the end task with the pre-training objective constructed from the end task data itself. In settings with few or no group annotations, we find that multitasking often, but not always, achieves better worst-group accuracy than Just-Train-Twice (JTT; Liu et al. (2021)) -- a representative distributionally robust optimization (DRO) method. Leveraging insights from synthetic data experiments, we propose to modify standard MTL by regularizing the joint multitask representation space. We run a large number of fine-tuning experiments across computer vision and natural language and find that our regularized MTL approach consistently outperforms JTT on both worst and average group outcomes. Our official code can be found here: .

cs

The demands on robotic manipulation skills to perform challenging tasks have drastically increased in recent times. To perform these tasks with dexterity, robots require perception tools to understand the scene and extract useful information that transforms to robot control inputs. To this end, recent research has introduced various object pose estimation and grasp pose detection methods that yield precise results. Assembly pose estimation is a secondary yet highly desirable skill in robotic assembling as it requires more detailed information on object placement as compared to bin picking and pick-and-place tasks. However, it has been often overlooked in research due to the complexity of integration in an agile framework. To address this issue, we propose an assembly pose estimation method with RGB-D input and 3D CAD models of the associated objects. The framework consists of semantic segmentation of the scene and registering point clouds of local surfaces against target point clouds derived from CAD models to estimate 6D poses. We show that our method can deliver sufficient accuracy for assembling object assemblies using evaluation metrics and demonstrations. The source code and dataset for the work can be found at:

cs

This paper presents a novel benchmarking framework Dyport for evaluating biomedical hypothesis generation systems. Utilizing curated datasets, our approach tests these systems under realistic conditions, enhancing the relevance of our evaluations. We integrate knowledge from the curated databases into a dynamic graph, accompanied by a method to quantify discovery importance. This not only assesses hypothesis accuracy but also their potential impact in biomedical research which significantly extends traditional link prediction benchmarks. Applicability of our benchmarking process is demonstrated on several link prediction systems applied on biomedical semantic knowledge graphs. Being flexible, our benchmarking system is designed for broad application in hypothesis generation quality verification, aiming to expand the scope of scientific discovery within the biomedical research community. Availability and implementation: Dyport framework is fully open-source. All code and datasets are available at:

cs

We prove that every paving matroid that is an excluded minor of interval positroids can be reduced to one of three fundamental families of excluded minors of interval positroids by relaxing dependent hyperplanes. Using this result, we classify all non-positroid excluded minors of interval positroids that are paving matroids. Additionally, we provide a criterion that characterizes all excluded minors of interval positroids that are paving positroids.

math

Electrons in materials containing heavy elements are fundamentally relativistic and should in principle be described using the Dirac equation. However, the current standard for treatment of electrons in such materials involves density functional theory methods originally formulated from the Schrödinger equation. While some extensions of the Schrödinger-based formulation have been explored, such as the scalar relativistic approximation with or without spin-orbit coupling, these solutions do not provide a way to fully account for all relativistic effects of electrons, and the language used to describe such solutions are still based in the language of the Schrödinger equation. In this article, we provide a different method for translating between the Dirac and Schrödinger viewpoints in the context of a Coulomb potential. By retaining the Dirac four-vector notation and terminology in taking the non-relativistic limit, we see a much deeper connection between the Dirac and Schrödinger equation solutions that allow us to more directly compare the effects of relativity in the angular and radial functions. Through this viewpoint, we introduce the concepts of densitals and Dirac spherical harmonics that allow us to translate more easily between the Dirac and Schrödinger solutions. These concepts allow us to establish a useful language for discussing relativistic effects in materials containing elements throughout the full periodic table and thereby enable a more fundamental understanding of the effects of relativity on electronic structure.

quant-ph

Opial's inequality and its ramifications play an important role in the theory of differential and difference equations. A sharp unifying generalization of Opial's inequality is presented that contains both its continuous and discrete version. This generalization based on distribution functions is extended to the case of derivatives of arbitrary order. This extension optimizes and improves the constant as given in the literature. The special case of derivatives of second order is studied in more detail. Two closely related Opial inequalities with a weight function are presented as well. The associated Wirtinger inequality is studied briefly.

math

This paper presents a novel approach to multi-agent reinforcement learning (RL) for linear systems with convex polytopic constraints. Existing works on RL has demonstrated the use of model predictive control (MPC) as a function approximator for the policy and value functions. The current paper is the first work to extend this idea to the multi-agent setting. We propose the use of a distributed MPC scheme as a function approximator, with a structure allowing for distributed learning and deployment, while maintaining the properties of centralized learning. The structured MPC scheme is introduced, and it is shown that the global policy and value functions can be approximated in a fully distributed manner. We then show that Q-learning updates can be performed distributively without introducing nonstationarity, by reconstructing a centralized learning update. The effectiveness of the approach is demonstrated on an academic example and a power systems case study.

eess

Exploring operational regimes of many-body cavity QED with multi-level atoms remains an exciting research frontier for their enhanced storage capabilities of intra-level quantum correlations. In this work, we propose an extension of a prototypical many-body cavity QED experiment from a two to a four-level description by optically addressing a combination of momentum and spin states of the ultracold atoms in the cavity. The resulting model comprises a pair of Dicke Hamiltonians constructed from pseudo-spin operators, effectively capturing two intertwined superradiant phase transitions. The phase diagram reveals regions featuring weak and strong entangled states of spin and momentum atomic degrees of freedom. These states exhibit different dynamical responses, ranging from slow to fast relaxation, with the added option of persistent entanglement temporal oscillations. We discuss the role of cavity losses in steering the system dynamics into such entangled states and propose a readout scheme that leverages different light polarizations within the cavity. Our work paves the way to connect the rich variety of non-equilibrium phase transitions that occur in many-body cavity QED to the buildup of quantum correlations in systems with multi-level atom descriptions.

quant-ph

The resolution is an important performance metric of near-field communication networks. In particular, the resolution of near field beamforming measures how effectively users can be distinguished in the distance-angle domain, which is one of the most significant features of near-field communications. In a comparison, conventional far-field beamforming can distinguish users in the angle domain only, which means that near-field communication yields the full utilization of user spatial resources to improve spectrum efficiency. In the literature of near-field communications, there have been a few studies on whether the resolution of near-field beamforming is perfect. However, each of the existing results suffers its own limitations, e.g., each is accurate for special cases only, and cannot precisely and comprehensively characterize the resolution. In this letter, a general analytical framework is developed to evaluate the resolution of near-field beamforming. Based on this derived expression, the impacts of parameters on the resolution are investigated, which can shed light on the design of the near-field communications, including the designs of beamforming and multiple access tequniques.

cs

Battery-limited devices and time-sensitive applications are considered as key players in the forthcoming wireless sensor network. Therefore, the main goal of the network is two-fold; Charge battery-limited devices, and provide status updates to users where information-freshness matters. In this paper, a multi-antenna base station (BS) in assistance of simultaneously-transmitting-and-reflecting reconfigurable intelligent surface (STAR-RIS) transmits power to energy-harvesting devices while controlling status update performance at information-users by analyzing age of information (AoI) metric. Therefore, we derive a scheduling policy at BS, and analyze joint transmit beamforming and amplitude-phase optimization at BS and STAR-RIS, respectively, to reduce average sum-AoI for the time-sensitive information-users while satisfying minimum required energy at energy-harvesting users. Moreover, two different energy-splitting and mode switching policies at STAR-RIS are studied. Then, by use of an alternating optimization algorithm, the optimization problem is studied and non-convexity of the problem is tackled by using the successive convex approximation technique. Through numerical results, AoI-metric and energy harvesting requirements of the network are analyzed versus different parameters such as number of antennas at BS, size of STAR-RIS, and transmitted power to highlight how we can improve two fold performance of the system by utilizing STAR-RIS compared to the conventional RIS structure.

math

Traffic congestion is a persistent problem in urban areas, which calls for the development of effective traffic signal control (TSC) systems. While existing Reinforcement Learning (RL)-based methods have shown promising performance in optimizing TSC, it is challenging to generalize these methods across intersections of different structures. In this work, a universal RL-based TSC framework is proposed for Vehicle-to-Everything (V2X) environments. The proposed framework introduces a novel agent design that incorporates a junction matrix to characterize intersection states, making the proposed model applicable to diverse intersections. To equip the proposed RL-based framework with enhanced capability of handling various intersection structures, novel traffic state augmentation methods are tailor-made for signal light control systems. Finally, extensive experimental results derived from multiple intersection configurations confirm the effectiveness of the proposed framework. The source code in this work is available at

cs

Latent space models are powerful statistical tools for modeling and understanding network data. While the importance of accounting for uncertainty in network analysis has been well recognized, the current literature predominantly focuses on point estimation and prediction, leaving the statistical inference of latent space models an open question. This work aims to fill this gap by providing a general framework to analyze the theoretical properties of the maximum likelihood estimators. In particular, we establish the uniform consistency and asymptotic distribution results for the latent space models under different edge types and link functions. Furthermore, the proposed framework enables us to generalize our results to the dependent-edge and sparse scenarios. Our theories are supported by simulation studies and have the potential to be applied in downstream inferences, such as link prediction and network testing problems.

stat

Tracking biosignals is crucial for monitoring wellness and preempting the development of severe medical conditions. Today, wearable devices can conveniently record various biosignals, creating the opportunity to monitor health status without disruption to one's daily routine. Despite widespread use of wearable devices and existing digital biomarkers, the absence of curated data with annotated medical labels hinders the development of new biomarkers to measure common health conditions. In fact, medical datasets are usually small in comparison to other domains, which is an obstacle for developing neural network models for biosignals. To address this challenge, we have employed self-supervised learning using the unlabeled sensor data collected under informed consent from the large longitudinal Apple Heart and Movement Study (AHMS) to train foundation models for two common biosignals: photoplethysmography (PPG) and electrocardiogram (ECG) recorded on Apple Watch. We curated PPG and ECG datasets from AHMS that include data from ~141K participants spanning ~3 years. Our self-supervised learning framework includes participant level positive pair selection, stochastic augmentation module and a regularized contrastive loss optimized with momentum training, and generalizes well to both PPG and ECG modalities. We show that the pre-trained foundation models readily encode information regarding participants' demographics and health conditions. To the best of our knowledge, this is the first study that builds foundation models using large-scale PPG and ECG data collected via wearable consumer devices $\unicode{x2013}$ prior works have commonly used smaller-size datasets collected in clinical and experimental settings. We believe PPG and ECG foundation models can enhance future wearable devices by reducing the reliance on labeled data and hold the potential to help the users improve their health.

eess

Magnitude predictions of $\Lambda$CDM, as parametrized by the Planck collaboration, are not consistent with the supernova data of the whole Pantheon+ sample even when, in order to take into account the uncertainty about its value, the Hubble constant is adjusted. This is a likely consequence of the increase of the number of low-redshift supernovae in the Pantheon+ sample, with respect to previous such samples. Indeed, when supernovae at red-shifts below 0.035 are ignored, $\Lambda$CDM predictions become consistent with Pantheon+ data. Interestingly, this is also the case if subsets of low-redshift supernovae roughly centered on the direction of the CMB dipole are considered, together with high-redshift ones. These results seem robust, since they are also obtained with a simple, single-parameter tired-light model.

astro-ph

Modern (sub)millimeter interferometers, such as ALMA and NOEMA, offer high angular resolution and unprecedented sensitivity. This provides the possibility to characterize the morphology of the gas and dust in distant galaxies. To assess the capabilities of current softwares in recovering morphologies and surface brightness profiles in interferometric observations, we test the performance of the Spergel model for fitting in the $uv$-plane, which has been recently implemented in the IRAM software GILDAS (uv$\_$fit). Spergel profiles provide an alternative to the Sersic profile, with the advantage of having an analytical Fourier transform, making them ideal to model visibilities in the $uv$-plane. We provide an approximate conversion between Spergel index and Sersic index, which depends on the ratio of the galaxy size to the angular resolution of the data. We show through extensive simulations that Spergel modeling in the $uv$-plane is a more reliable method for parameter estimation than modeling in the image-plane, as it returns parameters that are less affected by systematic biases and results in a higher effective signal-to-noise ratio (S/N). The better performance in the $uv$-plane is likely driven by the difficulty of accounting for correlated signal in interferometric images. Even in the $uv$-plane, the integrated source flux needs to be at least 50 times larger than the noise per beam to enable a reasonably good measurement of a Spergel index. We characterise the performance of Spergel model fitting in detail by showing that parameters biases are generally low (< 10%) and that uncertainties returned by uv$\_$fit are reliable within a factor of two. Finally, we showcase the power of Spergel fitting by re-examining two claims of extended halos around galaxies from the literature, showing that galaxies and halos can be successfully fitted simultaneously with a single Spergel model.

astro-ph

Ambient internet of things (IoT) is the network of devices which harvest energy from ambient sources for powering their communication. After decades of research on operation of these devices, Third Generation Partnership Project (3GPP) has started discussing energy harvesting technology in cellular networks to support massive deployment of IoT devices at low operational cost. This article provides a timely update on 3GPP studies on ambient energy harvesting devices including device types, use cases, key requirements, and related design challenges. Supported by link budget analysis for backscattering energy harvesting devices, which are a key component of this study, we provide insight on system design and show how this technology will require a new system design approach as compared to New Radio (NR) system design in 5G.

eess

Healthcare applications with the Internet of Things (IoT) are often safety-critical, thus, require extensive testing. Such applications are often connected to smart medical devices from various vendors. System-level testing of such applications requires test infrastructures physically integrating medical devices, which is time and monetary-wise expensive. Moreover, applications continuously evolve, e.g., introducing new devices and users and updating software. Nevertheless, a test infrastructure enabling testing with a few devices is insufficient for testing healthcare IoT systems, hence compromising their dependability. In this paper, we propose a model-based approach for the creation and operation of digital twins (DTs) of medicine dispensers as a replacement for physical devices to support the automated testing of IoT applications at scale. We evaluate our approach with an industrial IoT system with medicine dispensers in the context of Oslo City and its industrial partners, providing healthcare services to its residents. We study the fidelity of DTs in terms of their functional similarities with their physical counterparts: medicine dispensers. Results show that the DTs behave more than 92% similar to the physical medicine dispensers, providing a faithful replacement for the dispenser.

cs

In this paper we provide a different approach for existence of the variational solutions of the gradient flows associated to functionals on Sobolev spaces studied in \cite{BDDMS20}. The crucial condition is the convexity of the functional under which we show that the variational solutions coincide with the solutions generated by the nonlinear semigroup associated to the functional. For integral functionals of the form $\mathbf F(u)=\int_\Omega f(x,Du(x)) dx,$ where $f(x,\xi)$ is $C^1$ in $\xi$, we also make some remarks on the connections between convexity of $\mathbf F$ (called the integral convexity of $f$) and certain monotonicity conditions of the gradient map $D_\xi f.$ In particular, we provide an example to show that even for functions of the simple form $f=f(\xi)$, the usual quasimonotonicity of $D_\xi f$ is not sufficient for the integral convexity of $f.$

math

The Pólya group of an algebraic number field is the subgroup generated by the ideal classes of the products of prime ideals of equal norm inside the ideal class group. Inspired by a recent work on consecutive quadratic fields with large class numbers by Cherubini et al., we extend the notion of {\it consecutiveness} of number fields to certain parametric families of cyclic cubic fields and bi-quadratic fields and address the question of the existence of infinitely many such consecutive fields with large Pólya groups. This extends a recent result of the second author and Saikia for totally real bi-quadratic fields.

math

This work utilizes a variational autoencoder for channel estimation and evaluates it on real-world measurements. The estimator is trained solely on noisy channel observations and parameterizes an approximation to the mean squared error-optimal estimator by learning observation-dependent conditional first and second moments. The proposed estimator significantly outperforms related state-of-the-art estimators on real-world measurements. We investigate the effect of pre-training with synthetic data and find that the proposed estimator exhibits comparable results to the related estimators if trained on synthetic data and evaluated on the measurement data. Furthermore, pre-training on synthetic data also helps to reduce the required measurement training dataset size.

math

Battery-limited devices and time-sensitive applications are considered as key players in the forthcoming wireless sensor network. Therefore, the main goal of the network is two-fold; Charge battery-limited devices, and provide status updates to users where information-freshness matters. In this paper, a multi-antenna base station (BS) in assistance of simultaneously-transmitting-and-reflecting reconfigurable intelligent surface (STAR-RIS) transmits power to energy-harvesting devices while controlling status update performance at information-users by analyzing age of information (AoI) metric. Therefore, we derive a scheduling policy at BS, and analyze joint transmit beamforming and amplitude-phase optimization at BS and STAR-RIS, respectively, to reduce average sum-AoI for the time-sensitive information-users while satisfying minimum required energy at energy-harvesting users. Moreover, two different energy-splitting and mode switching policies at STAR-RIS are studied. Then, by use of an alternating optimization algorithm, the optimization problem is studied and non-convexity of the problem is tackled by using the successive convex approximation technique. Through numerical results, AoI-metric and energy harvesting requirements of the network are analyzed versus different parameters such as number of antennas at BS, size of STAR-RIS, and transmitted power to highlight how we can improve two fold performance of the system by utilizing STAR-RIS compared to the conventional RIS structure.

cs

Carlitz-compositions follow the restrictions of neighbouring parts $\sigma_{i-1}\neq\sigma_{i}$. The recently introduced Arndt-compositions have to satisfy $\sigma_{2i-1}>\sigma_{2i}$. The two concepts are combined to new and exciting objects that we call Arndt-Carlitz compositions.

math

With the growing prevalence of machine learning and artificial intelligence-based medical decision support systems, it is equally important to ensure that these systems provide patient outcomes in a fair and equitable fashion. This paper presents an innovative framework for detecting areas of algorithmic bias in medical-AI decision support systems. Our approach efficiently identifies potential biases in medical-AI models, specifically in the context of sepsis prediction, by employing the Classification and Regression Trees (CART) algorithm. We verify our methodology by conducting a series of synthetic data experiments, showcasing its ability to estimate areas of bias in controlled settings precisely. The effectiveness of the concept is further validated by experiments using electronic medical records from Grady Memorial Hospital in Atlanta, Georgia. These tests demonstrate the practical implementation of our strategy in a clinical environment, where it can function as a vital instrument for guaranteeing fairness and equity in AI-based medical decisions.

cs

Approximate message passing (AMP) has emerged both as a popular class of iterative algorithms and as a powerful analytic tool in a wide range of statistical estimation problems and statistical physics models. A well established line of AMP theory proves Gaussian approximations for the empirical distributions of the AMP iterate in the high dimensional limit, under the GOE random matrix model and its variants, via the celebrated conditioning technique. This paper provides a leave-one-out representation for the AMP iterate that holds under a broad class of Gaussian random matrix models with general variance profiles. In contrast to the existing AMP theory that describes the empirical distributions of the AMP iterate via a low dimensional state evolution, our leave-one-out representation yields an intrinsically high dimensional state evolution formula which characterizes the possibly heterogeneous, entrywise behavior of the AMP iterate under the prescribed random matrix models. To exemplify some distinct features of our AMP theory in applications, we analyze, in the context of regularized linear estimation, the precise stochastic behavior of the Ridge estimator for independent and non-identically distributed observations whose covariates exhibit general variance profiles. We find that its distribution is characterized via a weighted Ridge estimator in a heterogeneous Gaussian sequence model. Notably, in contrast to the i.i.d. sampling scenario, the effective noise and regularization are now full dimensional vectors determined via a high dimensional system of equations. Our method of proof differs significantly from the master conditioning approach, and relies on an inductive method that sheds light into the intricate cancellation scheme for the trace functional of certain random matrix recursion associated with the AMP.

cs

Gravitational $N$-body simulations calculate numerous interactions between particles. The tree algorithm reduces these calculations by constructing a hierarchical oct-tree structure and approximating gravitational forces on particles. Over the last three decades, the tree algorithm has been extensively used in large-scale simulations, and its parallelization in distributed memory environments has been well studied. However, recent supercomputers are equipped with many CPU cores per node, and optimizations of the tree construction in shared memory environments are becoming crucial. We propose a novel tree construction method in contrast to the conventional top-down approach. It first creates all leaf cells without traversing the tree and then constructs the remaining cells by a bottom-up approach. We evaluated the performance of our novel method on the supercomputer Fugaku and an Intel machine. On a single thread, our method accelerates one of the most time-consuming processes of the conventional tree construction method by a factor of above 3.0 on Fugaku and 2.2 on the Intel machine. Furthermore, as the number of threads increases, our parallel tree construction time reduces considerably. Compared to the conventional sequential tree construction method, we achieve a speedup of over 45 on 48 threads of Fugaku and more than 56 on 112 threads of the Intel machine. In stark contrast to the conventional method, the tree construction with our method no longer constitutes a bottleneck in the tree algorithm, even when using many threads.

astro-ph

Generating multi-view images from a single input view using image-conditioned diffusion models is a recent advancement and has shown considerable potential. However, issues such as the lack of consistency in synthesized views and over-smoothing in extracted geometry persist. Previous methods integrate multi-view consistency modules or impose additional supervisory to enhance view consistency while compromising on the flexibility of camera positioning and limiting the versatility of view synthesis. In this study, we consider the radiance field optimized during geometry extraction as a more rigid consistency prior, compared to volume and ray aggregation used in previous works. We further identify and rectify a critical bias in the traditional radiance field optimization process through score distillation from a multi-view diffuser. We introduce an Unbiased Score Distillation (USD) that utilizes unconditioned noises from a 2D diffusion model, greatly refining the radiance field fidelity. we leverage the rendered views from the optimized radiance field as the basis and develop a two-step specialization process of a 2D diffusion model, which is adept at conducting object-specific denoising and generating high-quality multi-view images. Finally, we recover faithful geometry and texture directly from the refined multi-view images. Empirical evaluations demonstrate that our optimized geometry and view distillation technique generates comparable results to the state-of-the-art models trained on extensive datasets, all while maintaining freedom in camera positioning.

cs

We introduce two lattice growth models: aggregation of $l$-dimensional boxes and aggregation of partitions with $l$ parts. We describe properties of the models: the parameter set of aggregations, the moments of the random variable of the number of growth directions, asymptotical behavior of proportions of the most frequent transitions of two- and three-dimensional self-aggregations.

math

Let $\mathcal{A}$ be a set of mutually coprime positive integers, satisfying \begin{align*} \sum\limits_{a\in\mathcal{A}}\frac{1}{a} = \infty. \end{align*} Define the (possibly non-multiplicative) "Liouville-like" functions \begin{align*} \lambda_{\mathcal{A}}(n) = (-1)^{\#\{a:a|n, a \in \mathcal{A}\}} \text{ or } (-1)^{\#\{a:a^\nu\parallel n, a \in \mathcal{A}, \nu \in \mathbb{N}\}}. \end{align*} We show that \begin{align*} \lim\limits_{x\to\infty}\frac{1}{x}\sum\limits_{n \leq x} \lambda_\mathcal{A}(n) = 0 \end{align*} holds, answering a question of de la Rue.

math

This paper introduces GeoShapley, a game theory approach to measuring spatial effects in machine learning models. GeoShapley extends the Nobel Prize-winning Shapley value framework in game theory by conceptualizing location as a player in a model prediction game, which enables the quantification of the importance of location and the synergies between location and other features in a model. GeoShapley is a model-agnostic approach and can be applied to statistical or black-box machine learning models in various structures. The interpretation of GeoShapley is directly linked with spatially varying coefficient models for explaining spatial effects and additive models for explaining non-spatial effects. Using simulated data, GeoShapley values are validated against known data-generating processes and are used for cross-comparison of seven statistical and machine learning models. An empirical example of house price modeling is used to illustrate GeoShapley's utility and interpretation with real world data. The method is available as an open-source Python package named geoshapley.

stat

Human educators possess an intrinsic ability to anticipate and seek educational explanations from students, which drives them to pose thought-provoking questions when students cannot articulate these explanations independently. We aim to imbue Intelligent Tutoring Systems with this ability using few-shot learning capability of Large Language Models. Our work proposes a novel prompting technique, Assertion Enhanced Few-Shot Learning, to facilitate the generation of accurate, detailed oriented educational explanations. Our central hypothesis is that, in educational domain, few-shot demonstrations are necessary but not a sufficient condition for quality explanation generation. We conducted a study involving 12 in-service teachers, comparing our approach to Traditional Few-Shot Learning. The results show that Assertion Enhanced Few-Shot Learning improves explanation accuracy by 15% and yields higher-quality explanations, as evaluated by teachers. We also conduct a qualitative ablation study to factor the impact of assertions to provide educator-friendly prompting guidelines for generating explanations in their domain of interest.

cs

Given an undirected graph G, the effective resistance r(s,t) measures the dissimilarity of node pair s,t in G, which finds numerous applications in real-world problems, such as recommender systems, combinatorial optimization, molecular chemistry, and electric power networks. Existing techniques towards pairwise effective resistance estimation either trade approximation guarantees for practical efficiency, or vice versa. In particular, the state-of-the-art solution is based on a multitude of Monte Carlo random walks, rendering it rather inefficient in practice, especially on large graphs. Motivated by this, this paper first presents an improved Monte Carlo approach, AMC, which reduces both the length and amount of random walks required without degrading the theoretical accuracy guarantee, through careful theoretical analysis and an adaptive sampling scheme. Further, we develop a greedy approach, GEER, which combines AMC with sparse matrix-vector multiplications in an optimized and non-trivial way. GEER offers significantly improved practical efficiency over AMC without compromising its asymptotic performance and accuracy guarantees. Extensive experiments on multiple benchmark datasets reveal that GEER is orders of magnitude faster than the state of the art in terms of computational time when achieving the same accuracy.

cs

A search is performed for charged-lepton flavor violating processes in top quark (t) production and decay. The data were collected by the CMS experiment from proton-proton collisions at a center-of-mass energy of 13 TeV and correspond to an integrated luminosity of 138 fb$^{-1}$. The selected events are required to contain one opposite-sign electron-muon pair, a third charged lepton (electron or muon), and at least one jet of which no more than one is associated with a bottom quark. Boosted decision trees are used to distinguish signal from background, exploiting differences in the kinematics of the final states particles. The data are consistent with the standard model expectation. Upper limits at 95% confidence level are placed in the context of effective field theory on the Wilson coefficients, which range between 0.024-0.424 TeV$^{-2}$ depending on the flavor of the associated light quark and the Lorentz structure of the interaction. These limits are converted to upper limits on branching fractions involving up (charm) quarks, t$\to$e$\mu$u (t$\to$e$\mu$c), of 0.032 (0.498)$\times$10$^{-6}$, 0.022 (0.369)$\times$10$^{-6}$, and 0.012 (0.216)$\times$10$^{-6}$ for tensor-like, vector-like, and scalar-like interactions, respectively.

hep-ex

In this paper, we explore the nature of influence in a network. The concept of participant-invariant influence is derived from an influence matrix M specifically designed to explore this phenomenon. Through nonnegative matrix factorization approximation, we managed to extract a participant-invariant matrix H representing a shared pattern that all participants must obey. The acquired H is highly field-related and can be further utilized to cluster factual networks. Our discovery of the unveiled participant-independent influence within network dynamics opens up new avenues for further research on network behavior and its implications.

cs

Diffusion models are a new class of generative models that have recently been applied to speech enhancement successfully. Previous works have demonstrated their superior performance in mismatched conditions compared to state-of-the art discriminative models. However, this was investigated with a single database for training and another one for testing, which makes the results highly dependent on the particular databases. Moreover, recent developments from the image generation literature remain largely unexplored for speech enhancement. These include several design aspects of diffusion models, such as the noise schedule or the reverse sampler. In this work, we systematically assess the generalization performance of a diffusion-based speech enhancement model by using multiple speech, noise and binaural room impulse response (BRIR) databases to simulate mismatched acoustic conditions. We also experiment with a noise schedule and a sampler that have not been applied to speech enhancement before. We show that the proposed system substantially benefits from using multiple databases for training, and achieves superior performance compared to state-of-the-art discriminative models in both matched and mismatched conditions. We also show that a Heun-based sampler achieves superior performance at a smaller computational cost compared to a sampler commonly used for speech enhancement.

cs

The macroscopic traffic flow model is widely used for traffic control and management. To incorporate drivers' anticipative behaviors and to remove impractical speed discontinuity inherent in the classic Lighthill-Whitham-Richards (LWR) traffic model, nonlocal partial differential equation (PDE) models with ``look-ahead" dynamics have been proposed, which assume that the speed is a function of weighted downstream traffic density. However, it lacks data validation on two important questions: whether there exist nonlocal dynamics, and how the length and weight of the ``look-ahead" window affect the spatial temporal propagation of traffic densities. In this paper, we adopt traffic trajectory data from a ring-road experiment and design a physics-informed neural network to learn the fundamental diagram and look-ahead kernel that best fit the data, and reinvent a data-enhanced nonlocal LWR model via minimizing the loss function combining the data discrepancy and the nonlocal model discrepancy. Results show that the learned nonlocal LWR yields a more accurate prediction of traffic wave propagation in three different scenarios: stop-and-go oscillations, congested, and free traffic. We first demonstrate the existence of ``look-ahead" effect with real traffic data. The optimal nonlocal kernel is found out to take a length of around 35 to 50 meters, and the kernel weight within 5 meters accounts for the majority of the nonlocal effect. Our results also underscore the importance of choosing a priori physics in machine learning models.

eess

Building taxonomies is often a significant part of building an ontology, and many attempts have been made to automate the creation of such taxonomies from relevant data. The idea in such approaches is either that relevant definitions of the intension of concepts can be extracted as patterns in the data (e.g. in formal concept analysis) or that their extension can be built from grouping data objects based on similarity (clustering). In both cases, the process leads to an automatically constructed structure, which can either be too coarse and lacking in definition, or too fined-grained and detailed, therefore requiring to be refined into the desired taxonomy. In this paper, we explore a method that takes inspiration from both approaches in an iterative and interactive process, so that refinement and definition of the concepts in the taxonomy occur at the time of identifying those concepts in the data. We show that this method is applicable on a variety of data sources and leads to taxonomies that can be more directly integrated into ontologies.

cs

We introduce an approach for augmenting text-to-video generation models with customized motions, extending their capabilities beyond the motions depicted in the original training data. By leveraging a few video samples demonstrating specific movements as input, our method learns and generalizes the input motion patterns for diverse, text-specified scenarios. Our contributions are threefold. First, to achieve our results, we finetune an existing text-to-video model to learn a novel mapping between the depicted motion in the input examples to a new unique token. To avoid overfitting to the new custom motion, we introduce an approach for regularization over videos. Second, by leveraging the motion priors in a pretrained model, our method can produce novel videos featuring multiple people doing the custom motion, and can invoke the motion in combination with other motions. Furthermore, our approach extends to the multimodal customization of motion and appearance of individualized subjects, enabling the generation of videos featuring unique characters and distinct motions. Third, to validate our method, we introduce an approach for quantitatively evaluating the learned custom motion and perform a systematic ablation study. We show that our method significantly outperforms prior appearance-based customization approaches when extended to the motion customization task.

cs

We study a planner's provision of information to manage workplace occupancy when strategic workers (agents) face risk of infectious disease transmission. The planner implements an information mechanism to signal information about the underlying risk of infection at the workplace. Agents update their belief over the risk parameter using this information and choose to work in-person or remotely. We address the design of the optimal signaling mechanism that best aligns the workplace occupancy with the planner's preference (i.e., maintaining safe capacity limits and operational efficiency at workplace). For various forms of planner preferences, we show numerical and analytical proof that interval-based information mechanisms are optimal. These mechanisms partition the continuous domain of the risk parameter into disjoint intervals and provision information based on interval-specific probability distributions over a finite set of signals. When the planner seeks to achieve an occupancy that lies in one of finitely many pre-specified ranges independent of the underlying risk, we provide an optimal mechanism that uses at most two intervals. On the other hand, when the preference on the occupancy is risk-dependent, we show that an approximately optimal interval-based mechanism can be computed efficiently. We bound the approximation loss for preferences that are expressed through a Lipschitz continuous function of both occupancy and risk parameter. We provide examples that demonstrate the improvement of proposed signaling mechanisms relative to the common benchmarks in information provision. Our findings suggest that information provision over the risk of disease transmission is an effective intervention for maintaining desirable occupancy levels at the workplace.

cs

In this paper, we give matrix formulae for non-orientable surfaces that provide the Laurent expansion for quasi-cluster variables, generalizing the orientable surface matrix formulae by Musiker-Williams. We additionally use our matrix formulas to prove the skein relations for the elements in the quasi-cluster algebra associated to curves on the non-orientable surface.

math

We explore the viability of casting foundation models as generic reward functions for reinforcement learning. To this end, we propose a simple pipeline that interfaces an off-the-shelf vision model with a large language model. Specifically, given a trajectory of observations, we infer the likelihood of an instruction describing the task that the user wants an agent to perform. We show that this generic likelihood function exhibits the characteristics ideally expected from a reward function: it associates high values with the desired behaviour and lower values for several similar, but incorrect policies. Overall, our work opens the possibility of designing open-ended agents for interactive tasks via foundation models.

cs

Classical snake robot control leverages mimicking snake-like gaits tuned for specific environments. However, to operate adaptively in unstructured environments, gait generation must be dynamically scheduled. In this work, we present a four-layer hierarchical control scheme to enable the snake robot to navigate freely in large-scale environments. The proposed model decomposes navigation into global planning, local planning, gait generation, and gait tracking. Using reinforcement learning (RL) and a central pattern generator (CPG), our method learns to navigate in complex mazes within hours and can be directly deployed to arbitrary new environments in a zero-shot fashion. We use the high-fidelity model of Northeastern's slithering robot COBRA to test the effectiveness of the proposed hierarchical control approach.

eess

Real-world datasets often exhibit imbalanced data distribution, where certain class levels are severely underrepresented. In such cases, traditional pattern classifiers have shown a bias towards the majority class, impeding accurate predictions for the minority class. This paper introduces an imbalanced data-oriented approach using probabilistic neural networks (PNNs) with a skew normal probability kernel to address this major challenge. PNNs are known for providing probabilistic outputs, enabling quantification of prediction confidence and uncertainty handling. By leveraging the skew normal distribution, which offers increased flexibility, particularly for imbalanced and non-symmetric data, our proposed Skew Probabilistic Neural Networks (SkewPNNs) can better represent underlying class densities. To optimize the performance of the proposed approach on imbalanced datasets, hyperparameter fine-tuning is imperative. To this end, we employ a population-based heuristic algorithm, Bat optimization algorithms, for effectively exploring the hyperparameter space. We also prove the statistical consistency of the density estimates which suggests that the true distribution will be approached smoothly as the sample size increases. Experimental simulations have been conducted on different synthetic datasets, comparing various benchmark-imbalanced learners. Our real-data analysis shows that SkewPNNs substantially outperform state-of-the-art machine learning methods for both balanced and imbalanced datasets in most experimental settings.

stat

Communication-efficient variants of SGD, specifically local SGD, have received a great deal of interest in recent years. These approaches compute multiple gradient steps locally, that is on each worker, before averaging model parameters, helping relieve the critical communication bottleneck in distributed deep learning training. Although many variants of these approaches have been proposed, they can sometimes lag behind state-of-the-art adaptive optimizers for deep learning. In this work, we investigate if the recent progress in the emerging area of learned optimizers can potentially close this gap while remaining communication-efficient. Specifically, we meta-learn how to perform global updates given an update from local SGD iterations. Our results demonstrate that learned optimizers can substantially outperform local SGD and its sophisticated variants while maintaining their communication efficiency. Learned optimizers can even generalize to unseen and much larger datasets and architectures, including ImageNet and ViTs, and to unseen modalities such as language modeling. We therefore demonstrate the potential of learned optimizers for improving communication-efficient distributed learning.

cs

We present the first measurement of the Weyl potential at four redshifts bins using data from the first three years of observations of the Dark Energy Survey (DES). The Weyl potential, which is the sum of the spatial and temporal distortions of the Universe's geometry, provides a direct way of testing the theory of gravity and the validity of the $\Lambda$CDM model. We find that the measured Weyl potential is 2.3$\sigma$, respectively 3.1$\sigma$, below the $\Lambda$CDM predictions in the two lowest redshift bins. We show that these low values of the Weyl potential are at the origin of the $\sigma_8$ tension between Cosmic Microwave Background (CMB) measurements and weak lensing measurements. Interestingly, we find that the tension remains if no information from the CMB is used. DES data on their own prefer a high value of the primordial fluctuations, followed by a slow evolution of the Weyl potential. A remarkable feature of our method is that the measurements of the Weyl potential are model-independent and can therefore be confronted with any theory of gravity, allowing efficient tests of models beyond General Relativity.

astro-ph

We prove a version of Shelah's Categoricity Conjecture for arbitrary deconstructible classes of modules. Moreover, we show that if $\mathcal{A}$ is a deconstructible class of modules closed under direct summands, and $\mathcal{A}$ fits in an abstract elementary class $(\mathcal{A},\preceq)$ such that $\preceq$ refines direct summands, then $\mathcal{A}$ is closed under arbitrary direct limits.

math

Resonant interactions associated with the emergence of a bound state constitute one of the cornerstones of modern many-body physics, ranging from Kondo physics, BEC-BCS crossover, to tunable interactions at Feshbach resonances in ultracold atoms or 2D semiconductors. Here we present a Feshbach perspective on the origin of strong pairing in Fermi-Hubbard type models. We perform a theoretical analysis of interactions between charge carriers in doped Mott insulators, modeled by a near-resonant two-channel scattering problem, and find strong evidence for Feshbach-type interactions in the $d_{x^2-y^2}$ channel that can support strong pairing, consistent with the established phenomenology of cuprates. Existing experimental and numerical results on hole-doped cuprates lead us to conjecture the existence of a light, long-lived, low-energy excited state of two holes with bipolaron character in these systems, which enables near-resonant interactions and can thus provide a microscopic foundation for theories of high-temperature superconductivity involving strong attraction, as assumed e.g. in BEC-BCS crossover scenarios. To put our theory to a direct test we suggest to use coincidence angle-resolved photoemission spectroscopy (cARPES), pair-tunneling measurements or less direct pump-probe experiments. The emergent Feshbach resonance we propose could also underlie superconductivity in other doped antiferromagnetic Mott insulators, as recently proposed for bilayer nickelates, highlighting its potential as a unifying strong-coupling pairing mechanism rooted in quantum magnetism.

quant-ph

Modulation recognition is a fundamental task in communication systems as the accurate identification of modulation schemes is essential for reliable signal processing, interference mitigation for coexistent communication technologies, and network optimization. Incorporating deep learning (DL) models into modulation recognition has demonstrated promising results in various scenarios. However, conventional DL models often fall short in online dynamic contexts, particularly in class incremental scenarios where new modulation schemes are encountered during online deployment. Retraining these models on all previously seen modulation schemes is not only time-consuming but may also not be feasible due to storage limitations. On the other hand, training solely on new modulation schemes often results in catastrophic forgetting of previously learned classes. This issue renders DL-based modulation recognition models inapplicable in real-world scenarios because the dynamic nature of communication systems necessitate the effective adaptability to new modulation schemes. This paper addresses this challenge by evaluating the performance of multiple Incremental Learning (IL) algorithms in dynamic modulation recognition scenarios, comparing them against conventional DL-based modulation recognition. Our results demonstrate that modulation recognition frameworks based on IL effectively prevent catastrophic forgetting, enabling models to perform robustly in dynamic scenarios.

eess

Hallucination is a well-known phenomenon in text generated by large language models (LLMs). The existence of hallucinatory responses is found in almost all application scenarios e.g., summarization, question-answering (QA) etc. For applications requiring high reliability (e.g., customer-facing assistants), the potential existence of hallucination in LLM-generated text is a critical problem. The amount of hallucination can be reduced by leveraging information retrieval to provide relevant background information to the LLM. However, LLMs can still generate hallucinatory content for various reasons (e.g., prioritizing its parametric knowledge over the context, failure to capture the relevant information from the context, etc.). Detecting hallucinations through automated methods is thus paramount. To facilitate research in this direction, we introduce a sophisticated dataset, DelucionQA, that captures hallucinations made by retrieval-augmented LLMs for a domain-specific QA task. Furthermore, we propose a set of hallucination detection methods to serve as baselines for future works from the research community. Analysis and case study are also provided to share valuable insights on hallucination phenomena in the target scenario.

cs

Artificial Neural Networks (ANN) have gained large popularity thanks to their ability to learn using the well-known backpropagation algorithm. On the other hand, Spiking Neural Networks (SNNs), despite having wider abilities than ANNs, have always presented a challenge in the training phase. This paper shows a new method to perform supervised learning on SNNs, using Spike Timing Dependent Plasticity (STDP) and homeostasis, aiming at training the network to identify spatial patterns. The method is tested using the SpiNNaker digital architecture. A SNN is trained to recognise one or multiple patterns and performance metrics are extracted to measure the performance of the network. Some considerations are drawn from the results showing that, in the case of a single trained pattern, the network behaves as the ideal detector, with 100% accuracy in detecting the trained pattern. However, as the number of trained patterns on a single network increases, the accuracy of the identification is linked to the similarities between these patterns. This method of training an SNN to detect spatial patterns may be applied on pattern recognition in static images or traffic analysis in computer networks, where each network packet represents a spatial pattern. It will be stipulated that the homeostatic factor may enable the network to detect patterns with some degree of similarities, rather than only perfectly matching patterns.

cs

We construct an Euler system for the adjoint Galois representation of a modular form, using motivic cohomology classes arising from Hilbert modular surfaces. We use this Euler system to give an upper bound for the Selmer group of the adjoint representation over the cyclotomic Zp-extension, which agrees with the predictions of the Iwasawa main conjecture up to powers of p and the augmentation ideal.

math

The reliability of wireless Ad Hoc Networks (WANET) communication is much lower than wired networks. WANET will be impacted by node overload, routing protocol, weather, obstacle blockage, and many other factors, all those anomalies cannot be avoided. Accurate prediction of the network entirely stopping in advance is essential after people could do networking re-routing or changing to different bands. In the present study, there are two primary goals. Firstly, design anomaly events detection patterns based on Metamorphic Testing (MT) methodology. Secondly, compare the performance of evaluation metrics, such as Transfer Rate, Occupancy rate, and the Number of packets received. Compared to other studies, the most significant advantage of mathematical interpretability, as well as not requiring dependence on physical environmental information, only relies on the networking physical layer and Mac layer data. The analysis of the results demonstrates that the proposed MT detection method is helpful for automatically identifying incidents/accident events on WANET. The physical layer transfer Rate metric could get the best performance.

stat

Deploying neural networks to different devices or platforms is in general challenging, especially when the model size is large or model complexity is high. Although there exist ways for model pruning or distillation, it is typically required to perform a full round of model training or finetuning procedure in order to obtain a smaller model that satisfies the model size or complexity constraints. Motivated by recent works on dynamic neural networks, we propose a simple way to train a large network and flexibly extract a subnetwork from it given a model size or complexity constraint during inference. We introduce a new way to allow a large model to be trained with dynamic depth and width during the training phase, and after the large model is trained we can select a subnetwork from it with arbitrary depth and width during the inference phase with a relatively better performance compared to training the subnetwork independently from scratch. Experiment results on a music source separation model show that our proposed method can effectively improve the separation performance across different subnetwork sizes and complexities with a single large model, and training the large model takes significantly shorter time than training all the different subnetworks.

eess

Human pose forecasting is the task of predicting articulated human motion given past human motion. There exists a number of popular benchmarks that evaluate an array of different models performing human pose forecasting. These benchmarks do not reflect that a human interacting system, such as a delivery robot, observes and plans for the motion of the same individual over an extended period of time. Every individual has unique and distinct movement patterns. This is however not reflected in existing benchmarks that evaluate a model's ability to predict an average human's motion rather than a particular individual's. We reformulate the human motion forecasting problem and present a model-agnostic personalization method. Motion forecasting personalization can be performed efficiently online by utilizing a low-parametric time-series analysis model that personalizes neural network pose predictions.

cs

Spatio-temporal forecasting in various domains, like traffic prediction and weather forecasting, is a challenging endeavor, primarily due to the difficulties in modeling propagation dynamics and capturing high-dimensional interactions among nodes. Despite the significant strides made by graph-based networks in spatio-temporal forecasting, there remain two pivotal factors closely related to forecasting performance that need further consideration: time delays in propagation dynamics and multi-scale high-dimensional interactions. In this work, we present a Series-Aligned Multi-Scale Graph Learning (SAMSGL) framework, aiming to enhance forecasting performance. In order to handle time delays in spatial interactions, we propose a series-aligned graph convolution layer to facilitate the aggregation of non-delayed graph signals, thereby mitigating the influence of time delays for the improvement in accuracy. To understand global and local spatio-temporal interactions, we develop a spatio-temporal architecture via multi-scale graph learning, which encompasses two essential components: multi-scale graph structure learning and graph-fully connected (Graph-FC) blocks. The multi-scale graph structure learning includes a global graph structure to learn both delayed and non-delayed node embeddings, as well as a local one to learn node variations influenced by neighboring factors. The Graph-FC blocks synergistically fuse spatial and temporal information to boost prediction accuracy. To evaluate the performance of SAMSGL, we conduct experiments on meteorological and traffic forecasting datasets, which demonstrate its effectiveness and superiority.

cs

In today's society, we are becoming increasingly dependent on software systems. However, we also constantly witness the negative impacts of buggy software. Program synthesis aims to improve software correctness by automatically generating the program given an outline of the expected behavior. For decades, program synthesis has been an active research field, with recent approaches looking to incorporate Large Language Models to help generate code. This paper explores the concept of LLM4TDD, where we guide Large Language Models to generate code iteratively using a test-driven development methodology. We conduct an empirical evaluation using ChatGPT and coding problems from LeetCode to investigate the impact of different test, prompt and problem attributes on the efficacy of LLM4TDD.

cs

Chest X-ray (CXR) is frequently employed in emergency departments and intensive care units to verify the proper placement of central lines and tubes and to rule out related complications. The automation of the X-ray reading process can be a valuable support tool for non-specialist technicians and minimize reporting delays due to non-availability of experts. While existing solutions for automated catheter segmentation and malposition detection show promising results, the disentanglement of individual catheters remains an open challenge, especially in complex cases where multiple devices appear superimposed in the X-ray projection. Moreover, conventional top-down instance segmentation methods are ineffective on such thin and long devices, that often extend through the entire image. In this paper, we propose a deep learning approach based on associative embeddings for catheter instance segmentation, able to overcome those limitations and effectively handle device intersections.

eess

Our understanding of the visual world is centered around various concept axes, characterizing different aspects of visual entities. While different concept axes can be easily specified by language, e.g. color, the exact visual nuances along each axis often exceed the limitations of linguistic articulations, e.g. a particular style of painting. In this work, our goal is to learn a language-informed visual concept representation, by simply distilling large pre-trained vision-language models. Specifically, we train a set of concept encoders to encode the information pertinent to a set of language-informed concept axes, with an objective of reproducing the input image through a pre-trained Text-to-Image (T2I) model. To encourage better disentanglement of different concept encoders, we anchor the concept embeddings to a set of text embeddings obtained from a pre-trained Visual Question Answering (VQA) model. At inference time, the model extracts concept embeddings along various axes from new test images, which can be remixed to generate images with novel compositions of visual concepts. With a lightweight test-time finetuning procedure, it can also generalize to novel concepts unseen at training.

cs

This paper proposes a data-driven framework to identify the attack-free sensors in a networked control system when some of the sensors are corrupted by an adversary. An operator with access to offline input-output attack-free trajectories of the plant is considered. Then, a data-driven algorithm is proposed to identify the attack-free sensors when the plant is controlled online. We also provide necessary conditions, based on the properties of the plant, under which the algorithm is feasible. An extension of the algorithm is presented to identify the sensors completely online against certain classes of attacks. The efficacy of our algorithm is depicted through numerical examples.

eess

In recent years there has been a dramatic increase in the number of malware attacks that use encrypted HTTP traffic for self-propagation or communication. Antivirus software and firewalls typically will not have access to encryption keys, and therefore direct detection of malicious encrypted data is unlikely to succeed. However, previous work has shown that traffic analysis can provide indications of malicious intent, even in cases where the underlying data remains encrypted. In this paper, we apply three machine learning techniques to the problem of distinguishing malicious encrypted HTTP traffic from benign encrypted traffic and obtain results comparable to previous work. We then consider the problem of feature analysis in some detail. Previous work has often relied on human expertise to determine the most useful and informative features in this problem domain. We demonstrate that such feature-related information can be obtained directly from machine learning models themselves. We argue that such a machine learning based approach to feature analysis is preferable, as it is more reliable, and we can, for example, uncover relatively unintuitive interactions between features.

cs

In this paper we revisit the classical problem of estimating a signal as it impinges on a multi-sensor array. We focus on the case where the impinging signal's bandwidth is appreciable and is operating in a broadband regime. Estimating broadband signals, often termed broadband (or wideband) beamforming, is traditionally done through filter and summation, true time delay, or a coupling of the two. Our proposed method deviates substantially from these paradigms in that it requires no notion of filtering or true time delay. We use blocks of samples taken directly from the sensor outputs to fit a robust Slepian subspace model using a least squares approach. We then leverage this model to estimate uniformly spaced samples of the impinging signal. Alongside a careful discussion of this model and how to choose its parameters we show how to fit the model to new blocks of samples as they are received, producing a streaming output. We then go on to show how this method naturally extends to adaptive beamforming scenarios, where we leverage signal statistics to attenuate interfering sources. Finally, we discuss how to use our model to estimate from dimensionality reducing measurements. Accompanying these discussions are extensive numerical experiments establishing that our method outperforms existing filter based approaches while being comparable in terms of computational complexity.

eess

Indirect experiments provide a valuable framework for estimating treatment effects in situations where conducting randomized control trials (RCTs) is impractical or unethical. Unlike RCTs, indirect experiments estimate treatment effects by leveraging (conditional) instrumental variables, enabling estimation through encouragement and recommendation rather than strict treatment assignment. However, the sample efficiency of such estimators depends not only on the inherent variability in outcomes but also on the varying compliance levels of users with the instrumental variables and the choice of estimator being used, especially when dealing with numerous instrumental variables. While adaptive experiment design has a rich literature for direct experiments, in this paper we take the initial steps towards enhancing sample efficiency for indirect experiments by adaptively designing a data collection policy over instrumental variables. Our main contribution is a practical computational procedure that utilizes influence functions to search for an optimal data collection policy, minimizing the mean-squared error of the desired (non-linear) estimator. Through experiments conducted in various domains inspired by real-world applications, we showcase how our method can significantly improve the sample efficiency of indirect experiments.

cs

We study the spreading of quantum information in a recently introduced family of brickwork quantum circuits that generalises the dual-unitary class. These circuits are unitary in time, while their spatial dynamics is unitary only in a restricted subspace. First, we show that local operators spread at the speed of light as in dual-unitary circuits, i.e., the butterfly velocity takes the maximal value allowed by the geometry of the circuit. Then, we prove that the entanglement spreading can still be characterised exactly for a family of compatible initial states (in fact, for an extension of the compatible family of dual-unitary circuits) and that the asymptotic entanglement slope is again independent on the Rényi index. Remarkably, however, we find that the entanglement velocity is generically smaller than one. We use these properties to find a closed-form expression for the entanglement membrane in these circuits.

quant-ph

A major challenge in monocular 3D object detection is the limited diversity and quantity of objects in real datasets. While augmenting real scenes with virtual objects holds promise to improve both the diversity and quantity of the objects, it remains elusive due to the lack of an effective 3D object insertion method in complex real captured scenes. In this work, we study augmenting complex real indoor scenes with virtual objects for monocular 3D object detection. The main challenge is to automatically identify plausible physical properties for virtual assets (e.g., locations, appearances, sizes, etc.) in cluttered real scenes. To address this challenge, we propose a physically plausible indoor 3D object insertion approach to automatically copy virtual objects and paste them into real scenes. The resulting objects in scenes have 3D bounding boxes with plausible physical locations and appearances. In particular, our method first identifies physically feasible locations and poses for the inserted objects to prevent collisions with the existing room layout. Subsequently, it estimates spatially-varying illumination for the insertion location, enabling the immersive blending of the virtual objects into the original scene with plausible appearances and cast shadows. We show that our augmentation method significantly improves existing monocular 3D object models and achieves state-of-the-art performance. For the first time, we demonstrate that a physically plausible 3D object insertion, serving as a generative data augmentation technique, can lead to significant improvements for discriminative downstream tasks such as monocular 3D object detection. Project website:

cs