Refined Analysis of Federated Averaging and Federated Richardson-Romberg

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Mangold, Paul, Durmus, Alain, Dieuleveut, Aymeric, Samsonov, Sergey, Moulines, Eric
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915685064507392
author Mangold, Paul
Durmus, Alain
Dieuleveut, Aymeric
Samsonov, Sergey
Moulines, Eric
author_facet Mangold, Paul
Durmus, Alain
Dieuleveut, Aymeric
Samsonov, Sergey
Moulines, Eric
contents In this paper, we present a novel analysis of \FedAvg with constant step size, relying on the Markov property of the underlying process. We demonstrate that the global iterates of the algorithm converge to a stationary distribution and analyze its resulting bias and variance relative to the problem's solution. We provide a first-order bias expansion in both homogeneous and heterogeneous settings. Interestingly, this bias decomposes into two distinct components: one that depends solely on stochastic gradient noise and another on client heterogeneity. Finally, we introduce a new algorithm based on the Richardson-Romberg extrapolation technique to mitigate this bias.
format Preprint
id arxiv_https___arxiv_org_abs_2412_01389
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Refined Analysis of Federated Averaging and Federated Richardson-Romberg
Mangold, Paul
Durmus, Alain
Dieuleveut, Aymeric
Samsonov, Sergey
Moulines, Eric
Machine Learning
Optimization and Control
In this paper, we present a novel analysis of \FedAvg with constant step size, relying on the Markov property of the underlying process. We demonstrate that the global iterates of the algorithm converge to a stationary distribution and analyze its resulting bias and variance relative to the problem's solution. We provide a first-order bias expansion in both homogeneous and heterogeneous settings. Interestingly, this bias decomposes into two distinct components: one that depends solely on stochastic gradient noise and another on client heterogeneity. Finally, we introduce a new algorithm based on the Richardson-Romberg extrapolation technique to mitigate this bias.
title Refined Analysis of Federated Averaging and Federated Richardson-Romberg
topic Machine Learning
Optimization and Control
url https://arxiv.org/abs/2412.01389