Learning-Augmented Decentralized Online Convex Optimization in Networks
This work addresses the problem of robust decentralized decision-making for networked agents, offering a novel approach that balances average performance with worst-case guarantees, though it builds incrementally on centralized learning-augmented methods.
The paper tackles decentralized online convex optimization in multi-agent networks by proposing the LADO algorithm, which uses a baseline policy for worst-case robustness and an ML policy for average performance, achieving strong robustness guarantees in decentralized settings and proving an average cost bound that shows a tradeoff between performance and robustness.
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.