Equilibria of Fully Decentralized Learning in Networked Systems

Yan Jiang, Wenqi Cui, Baosen Zhang, Jorge Cortes
Proceedings of The 5th Annual Learning for Dynamics and Control Conference, PMLR 211:333-345, 2023.

Abstract

Existing settings of decentralized learning either require players to have full information or the system to have certain special structure that may be hard to check and hinder their applicability to practical systems. To overcome this, we identify a structure that is simple to check for linear dynamical system, where each player learns in a fully decentralized fashion to minimize its cost. We first establish the existence of pure strategy Nash equilibria in the resulting noncooperative game. We then conjecture that the Nash equilibrium is unique provided that the system satisfies an additional requirement on its structure. We also introduce a decentralized mechanism based on projected gradient descent to have agents learn the Nash equilibrium. Simulations on a $5$-player game validate our results.

Cite this Paper

BibTeX
@InProceedings{pmlr-v211-jiang23a, title = {Equilibria of Fully Decentralized Learning in Networked Systems}, author = {Jiang, Yan and Cui, Wenqi and Zhang, Baosen and Cortes, Jorge}, booktitle = {Proceedings of The 5th Annual Learning for Dynamics and Control Conference}, pages = {333--345}, year = {2023}, editor = {Matni, Nikolai and Morari, Manfred and Pappas, George J.}, volume = {211}, series = {Proceedings of Machine Learning Research}, month = {15--16 Jun}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v211/jiang23a/jiang23a.pdf}, url = {https://proceedings.mlr.press/v211/jiang23a.html}, abstract = {Existing settings of decentralized learning either require players to have full information or the system to have certain special structure that may be hard to check and hinder their applicability to practical systems. To overcome this, we identify a structure that is simple to check for linear dynamical system, where each player learns in a fully decentralized fashion to minimize its cost. We first establish the existence of pure strategy Nash equilibria in the resulting noncooperative game. We then conjecture that the Nash equilibrium is unique provided that the system satisfies an additional requirement on its structure. We also introduce a decentralized mechanism based on projected gradient descent to have agents learn the Nash equilibrium. Simulations on a $5$-player game validate our results.} }
Endnote
%0 Conference Paper %T Equilibria of Fully Decentralized Learning in Networked Systems %A Yan Jiang %A Wenqi Cui %A Baosen Zhang %A Jorge Cortes %B Proceedings of The 5th Annual Learning for Dynamics and Control Conference %C Proceedings of Machine Learning Research %D 2023 %E Nikolai Matni %E Manfred Morari %E George J. Pappas %F pmlr-v211-jiang23a %I PMLR %P 333--345 %U https://proceedings.mlr.press/v211/jiang23a.html %V 211 %X Existing settings of decentralized learning either require players to have full information or the system to have certain special structure that may be hard to check and hinder their applicability to practical systems. To overcome this, we identify a structure that is simple to check for linear dynamical system, where each player learns in a fully decentralized fashion to minimize its cost. We first establish the existence of pure strategy Nash equilibria in the resulting noncooperative game. We then conjecture that the Nash equilibrium is unique provided that the system satisfies an additional requirement on its structure. We also introduce a decentralized mechanism based on projected gradient descent to have agents learn the Nash equilibrium. Simulations on a $5$-player game validate our results.
APA
Jiang, Y., Cui, W., Zhang, B. & Cortes, J.. (2023). Equilibria of Fully Decentralized Learning in Networked Systems. Proceedings of The 5th Annual Learning for Dynamics and Control Conference, in Proceedings of Machine Learning Research 211:333-345 Available from https://proceedings.mlr.press/v211/jiang23a.html.

Related Material