Accelerating Federated Edge Learning

doi:10.1109/LCOMM.2021.3103536

Accelerating Federated Edge Learning

2021-10-01

1

OA

AI

DOI：10.1109/LCOMM.2021.3103536

Original

Original request for help

Save

Abstract

En 中文

Transferring large models in federated learning (FL) networks is often hindered by clients' limited bandwidth. We propose FedAA, an FL algorithm which achieves fast convergence by exploiting the regularized Anderson acceleration (AA) on the global level. First, we demonstrate that FL can benefit from acceleration methods in numerical analysis. Second, FedAA improves the convergence rate for quadratic losses and improves the empirical performance for smooth and strongly convex objectives, compared to FedAvg, an FL algorithm using gradient descent (GD) local updates. Experimental results demonstrate that employing AA can significantly improve the performance of FedAvg, even when the objective is non-convex.

Keywords:

Convergence

Servers

Training

Urban areas

Numerical models

Collaborative work

Australia

Anderson acceleration

distributed optimization

federated learning

Accelerating Federated Edge Learning

Abstract

Journal

Organization