Mixed-strategy learning with continuous action sets

Abstract : Motivated by the recent applications of game-theoretical learning to the design of distributed control systems, we study a class of control problems that can be formulated as potential games with continuous action sets. We propose an actor-critic reinforcement learning algorithm that adapts mixed strategies over continuous action spaces. To analyze the algorithm, we extend the theory of finite-dimensional two-timescale stochastic approximation to a Banach space setting, and prove that the continuous dynamics of the process converge to equilibrium in the case of potential games. These results combine to give a provably-convergent learning algorithm in which players do not need to keep track of the controls selected by other agents.
Type de document :
Article dans une revue
IEEE Transactions on Automatic Control, Institute of Electrical and Electronics Engineers, 2017, 62 (1), pp.379 - 384. 〈10.1109/TAC.2015.2511930〉
Liste complète des métadonnées

https://hal.archives-ouvertes.fr/hal-01382280
Contributeur : Panayotis Mertikopoulos <>
Soumis le : dimanche 16 octobre 2016 - 15:29:01
Dernière modification le : vendredi 12 octobre 2018 - 01:18:06

Lien texte intégral

Identifiants

Citation

Steven Perkins, Panayotis Mertikopoulos, David S. Leslie. Mixed-strategy learning with continuous action sets. IEEE Transactions on Automatic Control, Institute of Electrical and Electronics Engineers, 2017, 62 (1), pp.379 - 384. 〈10.1109/TAC.2015.2511930〉. 〈hal-01382280〉

Partager

Métriques

Consultations de la notice

219