Convergence of Agnostic Federated Averaging

TL;DR

This paper provides the first convergence guarantees for agnostic Federated Averaging under realistic, non-uniform client participation, showing it outperforms other weighted aggregation methods in federated learning.

Contribution

It characterizes the convergence of agnostic FedAvg with stochastic, biased client participation without requiring knowledge of participation probabilities.

Findings

Achieves convergence rate of O(1/√T) for convex, nonsmooth losses.

First to analyze FedAvg under general stochastic client availability.

Agnostic FedAvg outperforms weighted variants even with known participation weights.

Abstract

Federated learning (FL) enables decentralized model training without centralizing raw data. However, practical FL deployments often face a key realistic challenge: Clients participate intermittently in server aggregation and with unknown, possibly biased participation probabilities. Most existing convergence results either assume full-device participation, or rely on knowledge of (in fact uniform) client availability distributions -- assumptions that rarely hold in practice. In this work, we characterize the optimization problem that consistently adheres to the stochastic dynamics of the well-known \emph{agnostic Federated Averaging (FedAvg)} algorithm under random (and variably-sized) client availability, and rigorously establish its convergence for convex, possibly nonsmooth losses, achieving a standard rate of order $\mathcal{O}(1/\sqrt{T})$, where $T$ denotes the aggregation horizon. Our analysis provides the first convergence guarantees for agnostic FedAvg under general, non-uniform, stochastic client participation, without knowledge of the participation distribution. We also empirically demonstrate that agnostic FedAvg in fact outperforms common (and suboptimal) weighted aggregation FedAvg variants, even with server-side knowledge of participation weights.