Saved in:
Bibliographic Details
Main Authors: Zhu, Shuchen, Huang, Zhengyang, Xu, Yuqi, Li, Peijin
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2604.25467
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913067896406016
author Zhu, Shuchen
Huang, Zhengyang
Xu, Yuqi
Li, Peijin
author_facet Zhu, Shuchen
Huang, Zhengyang
Xu, Yuqi
Li, Peijin
contents Federated learning increasingly operates in a large-model regime where communication, memory, and computation are all scarce. Typically, non-IID client data induce drift that degrades the stability and performance of local training. Existing remedies such as SCAFFOLD introduce heterogeneity-correction mechanisms to address this challenge, but they incur substantial extra communication and memory overhead. This paper proposes a subspace optimization method for federated learning (SSF), which performs heterogeneity-corrected optimization in a low-dimensional subspace using only projected quantities, while preserving full-dimensional control information through a backfill-style update that retains residual components whenever the active subspace changes. Under standard smoothness and bounded-variance assumptions, SSF attains a non-asymptotic rate of order $\widetilde{\mathcal{O}}(1/T+1/\sqrt{NKT})$. Experiments show favorable accuracy--efficiency trade-offs under heterogeneous data.
format Preprint
id arxiv_https___arxiv_org_abs_2604_25467
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Subspace Optimization for Efficient Federated Learning under Heterogeneous Data
Zhu, Shuchen
Huang, Zhengyang
Xu, Yuqi
Li, Peijin
Machine Learning
Optimization and Control
Federated learning increasingly operates in a large-model regime where communication, memory, and computation are all scarce. Typically, non-IID client data induce drift that degrades the stability and performance of local training. Existing remedies such as SCAFFOLD introduce heterogeneity-correction mechanisms to address this challenge, but they incur substantial extra communication and memory overhead. This paper proposes a subspace optimization method for federated learning (SSF), which performs heterogeneity-corrected optimization in a low-dimensional subspace using only projected quantities, while preserving full-dimensional control information through a backfill-style update that retains residual components whenever the active subspace changes. Under standard smoothness and bounded-variance assumptions, SSF attains a non-asymptotic rate of order $\widetilde{\mathcal{O}}(1/T+1/\sqrt{NKT})$. Experiments show favorable accuracy--efficiency trade-offs under heterogeneous data.
title Subspace Optimization for Efficient Federated Learning under Heterogeneous Data
topic Machine Learning
Optimization and Control
url https://arxiv.org/abs/2604.25467