Learning-Augmented Decentralized Online Convex Optimization in Networks

TL;DR

This paper introduces LADO, a decentralized online convex optimization algorithm that balances robustness and average performance by combining local online info with a machine learning policy in multi-agent networks.

Contribution

It presents the first learning-augmented decentralized algorithm with robustness guarantees, bridging the gap between centralized and decentralized online optimization.

Findings

LADO guarantees worst-case robustness in decentralized settings.

LADO improves average performance by leveraging ML policies.

Theoretical bounds demonstrate the tradeoff between robustness and average cost.

Abstract

This paper studies decentralized online convex optimization in a networked multi-agent system and proposes a novel algorithm, Learning-Augmented Decentralized Online optimization (LADO), for individual agents to select actions only based on local online information. LADO leverages a baseline policy to safeguard online actions for worst-case robustness guarantees, while staying close to the machine learning (ML) policy for average performance improvement. In stark contrast with the existing learning-augmented online algorithms that focus on centralized settings, LADO achieves strong robustness guarantees in a decentralized setting. We also prove the average cost bound for LADO, revealing the tradeoff between average performance and worst-case robustness and demonstrating the advantage of training the ML policy by explicitly considering the robustness requirement.