Exploratory LQG mean field games with entropy regularization

被引：10
作者：
Firoozi, Dena [1 ]
Jaimungal, Sebastian [2 ]
机构：
[1] HEC Montreal, Dept Decis Sci, Montreal, PQ, Canada
[2] Univ Toronto, Dept Stat Sci, Toronto, ON, Canada
来源：
AUTOMATICA | 2022年 / 139卷
基金：
加拿大自然科学与工程研究理事会;
关键词：
LQG mean field games; Entropy-regularization; Exploration;
D O I：
10.1016/j.automatica.2022.110177
中图分类号：
TP [自动化技术、计算机技术];
学科分类号：
0812 ;
摘要：
We study a general class of entropy-regularized multi-variate LQG mean field game (MFG) systems in continuous time with K distinct subpopulations of agents. We extend the notion of control actions to control distributions (exploratory actions), and explicitly derive the optimal control distributions for individual agents in the limiting MFG. We demonstrate that the optimal set of control distributions yields an epsilon-Nash equilibrium for the finite-population entropy-regularized MFG. Furthermore, we compare the resulting solutions with those of classical LQG MFG systems and establish the equivalence of their existence. (C)& nbsp;2022 Elsevier Ltd. All rights reserved.
引用
收藏
页数：12
相关论文
共 58 条
  • [51] Subramanian J, 2019, AAMAS '19: PROCEEDINGS OF THE 18TH INTERNATIONAL CONFERENCE ON AUTONOMOUS AGENTS AND MULTIAGENT SYSTEMS, P251
  • [52] Tenney Ian, 2014, IEEE INTCONF COMPUT
  • [53] Wang H., 2020, Journal of Machine Learning Research, V21, P1
  • [54] Wang Haoran, 2020, CONTINUOUS TIME MEAN
  • [55] Wang W., 2020, ARXIV PREPRINT ARXIV
  • [56] Yang J., 2017, ARXIV
  • [57] Learning in Mean-Field Games
    Yin, Huibing
    Mehta, Prashant G.
    Meyn, Sean P.
    Shanbhag, Uday V.
    [J]. IEEE TRANSACTIONS ON AUTOMATIC CONTROL, 2014, 59 (03) : 629 - 644
  • [58] Ziebart Brian D., 2008, P AAAI
123456