FedMDS: An Efficient Model Discrepancy-Aware Semi-Asynchronous Clustered Federated Learning Framework

Abstract

Federated learning (FL) is an emerging distributed machine learning paradigm that protects privacy and tackles the problem of isolated data islands. At present, there are two main communication strategies of FL: synchronous FL and asynchronous FL. The advantages of synchronous FL are the high precision and easy convergence of the model. However, this synchronous communication strategy has the risk of the straggler effect. Asynchronous FL has a natural advantage in mitigating the straggler effect, but there are threats of model quality degradation and server crash. In this paper, we propose a model discrepancy-aware semi-asynchronous clustered FL framework, <i>FedMDS</i> , which alleviates the straggler effect by 1) a clustered strategy based on the delay and direction of the model update and 2) a synchronous trigger mechanism that limits the model staleness. <i>FedMDS</i> leverages the clustered algorithm to reschedule the clients. Each group of clients performs asynchronous updates until the synchronous update mechanism based on the model discrepancy is triggered. We evaluate <i>FedMDS</i> based on four typical federated datasets in a non-IID setting and compare <i>FedMDS</i> to the baselines. The experimental results show that <i>FedMDS</i> significantly improves average test accuracy by more than <inline-formula><tex-math notation="LaTeX">$+9.2\%$</tex-math></inline-formula> on the four datasets compared to <i>TA-FedAvg</i> . In particular, <i>FedMDS</i> improves absolute Top-1 test accuracy by <inline-formula><tex-math notation="LaTeX">$+37.6\%$</tex-math></inline-formula> on FEMNIST compared to <i>TA-FedAvg</i> . The frequency of the average synchronization waiting time of <i>FedMDS</i> is significantly lower than that of <i>TA-FedAvg</i> on all datasets. Moreover, <i>FedMDS</i> can improve the accuracy and alleviate the straggler effect.