English

On the Performance of Gradient Tracking with Local Updates

Optimization and Control 2022-10-14 v2 Machine Learning Machine Learning

Abstract

We study the decentralized optimization problem where a network of nn agents seeks to minimize the average of a set of heterogeneous non-convex cost functions distributedly. State-of-the-art decentralized algorithms like Exact Diffusion~(ED) and Gradient Tracking~(GT) involve communicating every iteration. However, communication is expensive, resource intensive, and slow. In this work, we analyze a locally updated GT method (LU-GT), where agents perform local recursions before interacting with their neighbors. While local updates have been shown to reduce communication overhead in practice, their theoretical influence has not been fully characterized. We show LU-GT has the same communication complexity as the Federated Learning setting but allows arbitrary network topologies. In addition, we prove that the number of local updates does not degrade the quality of the solution achieved by LU-GT. Numerical examples reveal that local updates can lower communication costs in certain regimes (e.g., well-connected graphs).

Keywords

Cite

@article{arxiv.2210.04757,
  title  = {On the Performance of Gradient Tracking with Local Updates},
  author = {Edward Duc Hien Nguyen and Sulaiman A. Alghunaim and Kun Yuan and César A. Uribe},
  journal= {arXiv preprint arXiv:2210.04757},
  year   = {2022}
}

Comments

8 pages, 1 figure, submitted to ACC

R2 v1 2026-06-28T03:09:35.972Z