Distributed Online Randomized Gradient-Free Optimization With Compressed Communication

Longkang Zhu

doi:10.1109/tcyb.2026.3710570

Distributed Online Randomized Gradient-Free Optimization With Compressed Communication

2026-07-27

0

PRE

AI

DOI：10.1109/tcyb.2026.3710570

Original

Original request for help

Save

Abstract

En 中文

This article addresses two fundamental challenges in distributed online convex optimization (DOCO): communication efficiency and optimization under limited feedback. We propose a unified framework named online compressed gradient tracking (OCGT), which includes two variants: two-point bandit feedback (OCGT-BF) and stochastic gradient feedback (OCSGT). The proposed algorithms harness data compression and either gradient-free or stochastic gradient optimization techniques within distributed networks. The framework incorporates a compression scheme with error compensation mechanisms to reduce communication overhead while maintaining convergence guarantees. Unlike traditional approaches that assume perfect communication and full gradient access, OCGT operates effectively under practical constraints by combining gradient-like tracking with two-point or stochastic gradient feedback estimation. We provide a theoretical analysis demonstrating dynamic regret bounds for both variants. Finally, extensive experiments validate that OCGT achieves low dynamic regret while significantly reducing communication requirements.

Keywords:

Bandit feedback (BF)

communication compression

distributed online optimization

dynamic regret

gradient tracking

Distributed Online Randomized Gradient-Free Optimization With Compressed Communication

Abstract

Journal

Organization

Cited Papers