English

Averaging Rate Scheduler for Decentralized Learning on Heterogeneous Data

Machine Learning 2024-03-07 v1 Distributed, Parallel, and Cluster Computing

Abstract

State-of-the-art decentralized learning algorithms typically require the data distribution to be Independent and Identically Distributed (IID). However, in practical scenarios, the data distribution across the agents can have significant heterogeneity. In this work, we propose averaging rate scheduling as a simple yet effective way to reduce the impact of heterogeneity in decentralized learning. Our experiments illustrate the superiority of the proposed method (~3% improvement in test accuracy) compared to the conventional approach of employing a constant averaging rate.

Keywords

Cite

@article{arxiv.2403.03292,
  title  = {Averaging Rate Scheduler for Decentralized Learning on Heterogeneous Data},
  author = {Sai Aparna Aketi and Sakshi Choudhary and Kaushik Roy},
  journal= {arXiv preprint arXiv:2403.03292},
  year   = {2024}
}

Comments

9 pages, 3 figures, 4 tables. arXiv admin note: text overlap with arXiv:2305.04792

R2 v1 2026-06-28T15:10:20.258Z