Abstract:
Modern cyber-physical architectures use data col-lected from systems at different physical locations to learn appropriate behaviors and adapt to uncertain environments. H...Show MoreMetadata
Abstract:
Modern cyber-physical architectures use data col-lected from systems at different physical locations to learn appropriate behaviors and adapt to uncertain environments. However, an important challenge arises as communication exchanges at the edge of networked systems are costly due to limited resources. This paper considers a setup where multiple agents need to communicate efficiently in order to jointly solve a reinforcement learning problem over time-series data collected in a distributed manner. This is posed as learning an approximate value function over a communication network. An algorithm for achieving communication efficiency is proposed, supported with theoretical guarantees, practical implementations, and numerical evaluations. The approach is based on the idea of communicating only when sufficiently informative data is collected.
Published in: 2022 European Control Conference (ECC)
Date of Conference: 12-15 July 2022
Date Added to IEEE Xplore: 05 August 2022
ISBN Information: