Mean-Field Controls with Q-Learning for Cooperative MARL: Convergence and Complexity Analysis
Authors:
- Haotian GuHGHaotian Gu
- Xin GuoXGXin Guo
- Xiaoli WeiXWXiaoli Wei
- Renyuan XuRXRenyuan Xu
- 1 January 2021
- journal article
- research article
- Published by Society for Industrial & Applied Mathematics (SIAM) in SIAM Journal on Mathematics of Data Science
- Vol. 3 (4) , 1168-1196
Abstract
Multi-agent reinforcement learning (MARL), despite its popularity and empirical success, suffers from the curse of dimensionality. This paper builds the mathematical framework to approximate cooper...Keywords
All Related Versions
This publication has 28 references indexed in Scilit:
- A Stochastic Maximum Principle for Risk-Sensitive Mean-Field Type ControlIEEE Transactions on Automatic Control, 2015
- Mean Field Equilibria of Dynamic Auctions with LearningManagement Science, 2014
- Large-Population Cost-Coupled LQG Problems With Nonuniform Agents: Individual-Mass Behavior and Decentralized $\varepsilon$-Nash EquilibriaIEEE Transactions on Automatic Control, 2007
- Mean field gamesJapanese Journal of Mathematics, 2007
- Large population stochastic dynamic games: closed-loop McKean-Vlasov systems and the Nash certainty equivalence principleCommunications in Information and Systems, 2006
- Lipschitz Continuity of Value Functions in Markovian Decision ProcessesMathematical Methods of Operations Research, 2005
- Kernel-based reinforcement learning in average-cost problemsIEEE Transactions on Automatic Control, 2002
- On Choosing and Bounding Probability MetricsInternational Statistical Review, 2002
- Algorithms for optimization and stabilization of controlled Markov chainsSādhanā, 1999
- On the McKean‐Vlasov Limit for Interacting DiffusionsMathematische Nachrichten, 1988