AI Paper Atlas

Research map / Multi-agent systems and games / Cooperative multi-agent RL

A Multi-Agent Reinforcement Learning Approach for Efficient Client Selection in Federated Learning

Sai Qian Zhang, Jieyu Lin, Qi Zhang

AAAI 2022 · Multi-agent systems and games › Cooperative multi-agent RL · approach: MARL benchmarks and grouping · 115 citations (OpenAlex)

Read the paper

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Use MARL for runtime client selection based on probing loss and system statisticsFederated learning · MNIST, CIFAR-10, Fashion MNIST, Shakespearetest accuracy: 96.91% vs 96.86%; 48.87% vs 48.19%; 96.14% vs 96.08%; 44.58% vs 44.10%Improved
Use probing loss to select clients, rejecting those with above-average probing lossFedAvg under non-IID client data · MNIST and CIFAR-10test accuracy: 98.7% and 45.2%; 1.3% and 2.8% higher than the other methods on averageImproved
Use multi-agent reinforcement learning rather than single-agent reinforcement learningFedMarl · Federated learning training environmentconvergence speedImproved
Increase latency and communication-cost reward weights from 0.2 and 0.1 to 0.5 and 0.3 (ablation)FedMarl · LeNet, VGG6, ResNet-18processing latency: 1.0× -> 0.82×; 1.0× -> 0.80×; 1.0× -> 0.80×Improved
Increase latency and communication-cost reward weights from 0.2 and 0.1 to 0.5 and 0.3 (ablation)FedMarl · LeNet, VGG6, ResNet-18communication cost: 1.0× -> 0.83×; 1.0× -> 0.78×; 1.0× -> 0.86×Improved
Increase latency and communication-cost reward weights from 0.2 and 0.1 to 0.5 and 0.3 (ablation)FedMarl · LeNet, VGG6, ResNet-18test accuracy: 98.15% -> 97.88%; 46.05% -> 45.28%; 98.43% -> 98.03%Hurt
Use agents trained on another DNN and dataset without retrainingFedMarl · LeNet on MNISTnormalized reward: 0.965 (0.989)Mixed
Use normalized agent-state inputs to transfer trained agents to different client-pool sizesFedMarl · CIFAR-10, K=400 and K=800total reward: 1.36× and 1.32× higher ... on averageImproved
Use normalized agent-state inputs to transfer trained agents to a pool with new device typesFedMarl · CIFAR-10 with six additional mobile device typesoverall performanceImproved
Reduce processing latency and communication cost through FedMarl client selectionFederated learning · Evaluation benchmarksprocessing latency: 1.7×Improved
Reduce processing latency and communication cost through FedMarl client selectionFederated learning · Evaluation benchmarkscommunication cost: 2.2×Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy