Learning from straggler clients in federated learning

TL;DR

This paper investigates the challenge of learning from delayed client updates in federated learning, proposing new algorithms that improve accuracy and efficiency when dealing with straggler clients.

Contribution

It introduces two novel algorithms, FARe-DUST and FeAST-on-MSG, designed to better handle severely delayed client updates in federated learning.

Findings

New algorithms outperform existing methods in accuracy for delayed clients.

Proposed methods achieve better trade-offs between training time and overall accuracy.

Experiments on multiple benchmarks validate the effectiveness of the new approaches.

Abstract

How well do existing federated learning algorithms learn from client devices that return model updates with a significant time delay? Is it even possible to learn effectively from clients that report back minutes, hours, or days after being scheduled? We answer these questions by developing Monte Carlo simulations of client latency that are guided by real-world applications. We study synchronous optimization algorithms like FedAvg and FedAdam as well as the asynchronous FedBuff algorithm, and observe that all these existing approaches struggle to learn from severely delayed clients. To improve upon this situation, we experiment with modifications, including distillation regularization and exponential moving averages of model weights. Finally, we introduce two new algorithms, FARe-DUST and FeAST-on-MSG, based on distillation and averaging, respectively. Experiments with the EMNIST, CIFAR-100, and StackOverflow benchmark federated learning tasks demonstrate that our new algorithms outperform existing ones in terms of accuracy for straggler clients, while also providing better trade-offs between training time and total accuracy.