FedSTaS: Client Stratification and Client Level Sampling for Efficient Federated Learning

TL;DR

FedSTaS introduces a novel client and data sampling method for federated learning that improves accuracy and efficiency by stratifying clients and optimally sampling data, addressing communication and privacy challenges.

Contribution

It proposes FedSTaS, a new sampling approach that enhances federated learning efficiency by client stratification and optimal data sampling, outperforming previous methods.

Findings

FedSTaS achieves higher accuracy than FedSTS.

It reduces the number of training rounds needed.

Experiments on three datasets validate its effectiveness.

Abstract

Federated learning (FL) is a machine learning methodology that involves the collaborative training of a global model across multiple decentralized clients in a privacy-preserving way. Several FL methods are introduced to tackle communication inefficiencies but do not address how to sample participating clients in each round effectively and in a privacy-preserving manner. In this paper, we propose \textit{FedSTaS}, a client and data-level sampling method inspired by \textit{FedSTS} and \textit{FedSampling}. In each federated learning round, \textit{FedSTaS} stratifies clients based on their compressed gradients, re-allocate the number of clients to sample using an optimal Neyman allocation, and sample local data from each participating clients using a data uniform sampling strategy. Experiments on three datasets show that \textit{FedSTaS} can achieve higher accuracy scores than those of \textit{FedSTS} within a fixed number of training rounds.