Communication-Efficient Heterogeneous Federated Learning with Generalized Heavy-Ball Momentum
Keywords: federated learning, momentum, distributed learning, deep learning
TL;DR: A novel formulation for momentum that mitigates the issues of statistical heterogeneity in FL, speed up convergence and improve communication efficiency in realistic large-scale scenarios.
Abstract: Federated Learning (FL) has emerged as the state-of-the-art approach for learning from decentralized data in privacy-constrained scenarios. However, system and statistical challenges hinder real-world applications, which demand efficient learning from edge devices and robustness to heterogeneity. Despite significant research efforts, existing approaches (i) are not sufficiently robust, (ii) do not perform well in large-scale scenarios, and (iii) are not communication efficient. In this work, we propose a novel _Generalized Heavy-Ball Momentum_ (GHBM), proving that it enjoys an improved theoretical convergence rate w.r.t. existing FL methods based on classical momentum in _partial participation_, without relying on bounded data heterogeneity. Then, we present FedHBM as an adaptive, communication-efficient by-design instance of GHBM. Extensive experimentation on vision and language tasks, in both controlled and realistic large-scale scenarios, confirms our theoretical findings, showing that GHBM substantially improves the state of the art, especially in large scale scenarios with high data heterogeneity and low client participation.
Supplementary Material: zip
Primary Area: optimization
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.
Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.
Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.
No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.
Submission Number: 6420
Loading