English

The Min-Max Complexity of Distributed Stochastic Convex Optimization with Intermittent Communication

Machine Learning 2021-08-06 v2 Optimization and Control

Abstract

We resolve the min-max complexity of distributed stochastic convex optimization (up to a log factor) in the intermittent communication setting, where MM machines work in parallel over the course of RR rounds of communication to optimize the objective, and during each round of communication, each machine may sequentially compute KK stochastic gradient estimates. We present a novel lower bound with a matching upper bound that establishes an optimal algorithm.

Keywords

Cite

@article{arxiv.2102.01583,
  title  = {The Min-Max Complexity of Distributed Stochastic Convex Optimization with Intermittent Communication},
  author = {Blake Woodworth and Brian Bullins and Ohad Shamir and Nathan Srebro},
  journal= {arXiv preprint arXiv:2102.01583},
  year   = {2021}
}

Comments

48 pages

R2 v1 2026-06-23T22:46:12.030Z