PFLlib/system/flcore/servers/servercp.py

# PFLlib: Personalized Federated Learning Algorithm Library
# Copyright (C) 2021  Jianqing Zhang

# This program is free software; you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation; either version 2 of the License, or
# (at your option) any later version.

# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
# GNU General Public License for more details.

# You should have received a copy of the GNU General Public License along
# with this program; if not, write to the Free Software Foundation, Inc.,
# 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.

import copy
import torch
import time
from flcore.clients.clientcp import *
from flcore.servers.serverbase import Server
from utils.data_utils import read_client_data
from threading import Thread


class FedCP(Server):
    def __init__(self, args, times):
        super().__init__(args, times)

        in_dim = list(args.model.head.parameters())[0].shape[1]
        cs = ConditionalSelection(in_dim, in_dim).to(args.device)

        # select slow clients
        self.set_slow_clients()
        for i, train_slow, send_slow in zip(range(self.num_clients), self.train_slow_clients, self.send_slow_clients):
            train_data = read_client_data(self.dataset, i, is_train=True)
            test_data = read_client_data(self.dataset, i, is_train=False)
            client = clientCP(self.args, 
                            id=i, 
                            train_samples=len(train_data), 
                            test_samples=len(test_data), 
                            train_slow=train_slow, 
                            send_slow=send_slow,
                            ConditionalSelection=cs)
            self.clients.append(client)

        print(f"\nJoin ratio / total clients: {self.join_ratio} / {self.num_clients}")
        print("Finished creating server and clients.")

        # self.load_model()
        self.Budget = []
        self.head = None
        self.cs = None


    def send_models(self):
        assert (len(self.clients) > 0)

        for client in self.clients:
            client.set_parameters(self.global_modules)

    def add_parameters(self, w, client_model):
        for server_param, client_param in zip(self.global_modules.parameters(), client_model.parameters()):
            server_param.data += client_param.data.clone() * w

    def aggregate_parameters(self):
        assert (len(self.uploaded_models) > 0)

        self.global_modules = copy.deepcopy(self.uploaded_models[0])
        for param in self.global_modules.parameters():
            param.data = torch.zeros_like(param.data)
            
        for w, client_model in zip(self.uploaded_weights, self.uploaded_models):
            self.add_parameters(w, client_model)

    def evaluate(self, acc=None):
        stats = self.test_metrics()

        test_acc = sum(stats[2])*1.0 / sum(stats[1])
        test_auc = sum(stats[3])*1.0 / sum(stats[1])
        
        if acc == None:
            self.rs_test_acc.append(test_acc)
        else:
            acc.append(test_acc)

        print("Averaged Test Accurancy: {:.4f}".format(test_acc))
        print("Averaged Test AUC: {:.4f}".format(test_auc))


    def train(self):
        for i in range(self.global_rounds+1):
            s_t = time.time()
            self.selected_clients = self.select_clients()

            if i%self.eval_gap == 0:
                print(f"\n-------------Round number: {i}-------------")
                print("\nEvaluate before local training")
                self.evaluate()

            for client in self.selected_clients:
                client.train_cs_model()
                client.generate_upload_head()

            self.receive_models()
            self.aggregate_parameters()
            self.send_models()
            self.global_head()
            self.global_cs()

            self.Budget.append(time.time() - s_t)
            print('-'*50, self.Budget[-1])

        print("\nBest global accuracy.")
        print(max(self.rs_test_acc))
        print("\nAverage time cost per round.")
        print(sum(self.Budget[1:])/len(self.Budget[1:]))


    def receive_models(self):
        assert (len(self.selected_clients) > 0)

        active_train_samples = 0
        for client in self.selected_clients:
            active_train_samples += client.train_samples

        self.uploaded_weights = []
        self.uploaded_ids = []
        self.uploaded_models = []
        for client in self.selected_clients:
            self.uploaded_weights.append(client.train_samples / active_train_samples)
            self.uploaded_ids.append(client.id)
            self.uploaded_models.append(client.model.model.base)

    def global_head(self):
        self.uploaded_model_gs = []
        for client in self.selected_clients:
            self.uploaded_model_gs.append(client.model.head_g)

        self.head = copy.deepcopy(self.uploaded_model_gs[0])
        for param in self.head.parameters():
            param.data = torch.zeros_like(param.data)
            
        for w, client_model in zip(self.uploaded_weights, self.uploaded_model_gs):
            self.add_head(w, client_model)

        for client in self.selected_clients:
            client.set_head_g(self.head)

    def add_head(self, w, head):
        for server_param, client_param in zip(self.head.parameters(), head.parameters()):
            server_param.data += client_param.data.clone() * w
            
    def global_cs(self):
        self.uploaded_model_gs = []
        for client in self.selected_clients:
            self.uploaded_model_gs.append(client.model.gate.cs)

        self.cs = copy.deepcopy(self.uploaded_model_gs[0])
        for param in self.cs.parameters():
            param.data = torch.zeros_like(param.data)
            
        for w, client_model in zip(self.uploaded_weights, self.uploaded_model_gs):
            self.add_cs(w, client_model)

        for client in self.selected_clients:
            client.set_cs(self.cs)

    def add_cs(self, w, cs):
        for server_param, client_param in zip(self.cs.parameters(), cs.parameters()):
            server_param.data += client_param.data.clone() * w


class ConditionalSelection(nn.Module):
    def __init__(self, in_dim, h_dim):
        super(ConditionalSelection, self).__init__()
        
        self.fc = nn.Sequential(
            nn.Linear(in_dim, h_dim*2), 
            nn.LayerNorm([h_dim*2]),
            nn.ReLU(),
        )

    def forward(self, x, tau=1, hard=False):
        shape = x.shape
        x = self.fc(x)
        x = x.view(shape[0], 2, -1)
        x = F.gumbel_softmax(x, dim=1, tau=tau, hard=hard)
        return x[:, 0, :], x[:, 1, :]