FedGiA: An Efficient Hybrid Algorithm for Federated Learning

TL;DR

FedGiA is a hybrid federated learning algorithm that improves communication efficiency, reduces computational costs, and guarantees convergence, addressing key challenges in federated learning.

Contribution

The paper introduces FedGiA, a novel hybrid algorithm combining gradient descent and ADMM, with proven global convergence and enhanced efficiency.

Findings

FedGiA outperforms existing algorithms in communication and computation efficiency.

FedGiA demonstrates global convergence under mild conditions.

Numerical experiments confirm the theoretical advantages of FedGiA.

Abstract

Federated learning has shown its advances recently but is still facing many challenges, such as how algorithms save communication resources and reduce computational costs, and whether they converge. To address these critical issues, we propose a hybrid federated learning algorithm (FedGiA) that combines the gradient descent and the inexact alternating direction method of multipliers. The proposed algorithm is more communication- and computation-efficient than several state-of-the-art algorithms theoretically and numerically. Moreover, it also converges globally under mild conditions.