Beyond Local Sharpness: Communication-Efficient Global Sharpness-aware Minimization for Federated Learning

TL;DR

FedGloSS is a federated learning method that optimizes global sharpness on the server using a communication-efficient approximation, leading to flatter minima and improved model performance.

Contribution

Introduces FedGloSS, a novel server-side sharpness-aware minimization approach that reduces communication costs while enhancing global model flatness in federated learning.

Findings

FedGloSS achieves flatter minima than existing methods.

It outperforms state-of-the-art FL approaches on vision benchmarks.

The method reduces communication overhead significantly.

Abstract

Federated learning (FL) enables collaborative model training with privacy preservation. Data heterogeneity across edge devices (clients) can cause models to converge to sharp minima, negatively impacting generalization and robustness. Recent approaches use client-side sharpness-aware minimization (SAM) to encourage flatter minima, but the discrepancy between local and global loss landscapes often undermines their effectiveness, as optimizing for local sharpness does not ensure global flatness. This work introduces FedGloSS (Federated Global Server-side Sharpness), a novel FL approach that prioritizes the optimization of global sharpness on the server, using SAM. To reduce communication overhead, FedGloSS cleverly approximates sharpness using the previous global gradient, eliminating the need for additional client communication. Our extensive evaluations demonstrate that FedGloSS consistently reaches flatter minima and better performance compared to state-of-the-art FL methods across various federated vision benchmarks.