Maximum Entropy Correlated Equilibria

Luis E. Ortiz, Robert E. Schapire, Sham M. Kakade
; Proceedings of the Eleventh International Conference on Artificial Intelligence and Statistics, PMLR 2:347-354, 2007.

Abstract

We study maximum entropy correlated equilibria (Maxent CE) in multi-player games. After motivating and deriving some interesting important properties of Maxent CE, we provide two gradient-based algorithms that are guaranteed to converge to it. The proposed algorithms have strong connections to algorithms for statistical estimation (e.g., iterative scaling), and permit a distributed learning-dynamics interpretation. We also briefly discuss possible connections of this work, and more generally of the Maximum Entropy Principle in statistics, to the work on learning in games and the problem of equilibrium selection.

Cite this Paper


BibTeX
@InProceedings{pmlr-v2-ortiz07a, title = {Maximum Entropy Correlated Equilibria}, author = {Luis E. Ortiz and Robert E. Schapire and Sham M. Kakade}, booktitle = {Proceedings of the Eleventh International Conference on Artificial Intelligence and Statistics}, pages = {347--354}, year = {2007}, editor = {Marina Meila and Xiaotong Shen}, volume = {2}, series = {Proceedings of Machine Learning Research}, address = {San Juan, Puerto Rico}, month = {21--24 Mar}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v2/ortiz07a/ortiz07a.pdf}, url = {http://proceedings.mlr.press/v2/ortiz07a.html}, abstract = {We study maximum entropy correlated equilibria (Maxent CE) in multi-player games. After motivating and deriving some interesting important properties of Maxent CE, we provide two gradient-based algorithms that are guaranteed to converge to it. The proposed algorithms have strong connections to algorithms for statistical estimation (e.g., iterative scaling), and permit a distributed learning-dynamics interpretation. We also briefly discuss possible connections of this work, and more generally of the Maximum Entropy Principle in statistics, to the work on learning in games and the problem of equilibrium selection.} }
Endnote
%0 Conference Paper %T Maximum Entropy Correlated Equilibria %A Luis E. Ortiz %A Robert E. Schapire %A Sham M. Kakade %B Proceedings of the Eleventh International Conference on Artificial Intelligence and Statistics %C Proceedings of Machine Learning Research %D 2007 %E Marina Meila %E Xiaotong Shen %F pmlr-v2-ortiz07a %I PMLR %J Proceedings of Machine Learning Research %P 347--354 %U http://proceedings.mlr.press %V 2 %W PMLR %X We study maximum entropy correlated equilibria (Maxent CE) in multi-player games. After motivating and deriving some interesting important properties of Maxent CE, we provide two gradient-based algorithms that are guaranteed to converge to it. The proposed algorithms have strong connections to algorithms for statistical estimation (e.g., iterative scaling), and permit a distributed learning-dynamics interpretation. We also briefly discuss possible connections of this work, and more generally of the Maximum Entropy Principle in statistics, to the work on learning in games and the problem of equilibrium selection.
RIS
TY - CPAPER TI - Maximum Entropy Correlated Equilibria AU - Luis E. Ortiz AU - Robert E. Schapire AU - Sham M. Kakade BT - Proceedings of the Eleventh International Conference on Artificial Intelligence and Statistics PY - 2007/03/11 DA - 2007/03/11 ED - Marina Meila ED - Xiaotong Shen ID - pmlr-v2-ortiz07a PB - PMLR SP - 347 DP - PMLR EP - 354 L1 - http://proceedings.mlr.press/v2/ortiz07a/ortiz07a.pdf UR - http://proceedings.mlr.press/v2/ortiz07a.html AB - We study maximum entropy correlated equilibria (Maxent CE) in multi-player games. After motivating and deriving some interesting important properties of Maxent CE, we provide two gradient-based algorithms that are guaranteed to converge to it. The proposed algorithms have strong connections to algorithms for statistical estimation (e.g., iterative scaling), and permit a distributed learning-dynamics interpretation. We also briefly discuss possible connections of this work, and more generally of the Maximum Entropy Principle in statistics, to the work on learning in games and the problem of equilibrium selection. ER -
APA
Ortiz, L.E., Schapire, R.E. & Kakade, S.M.. (2007). Maximum Entropy Correlated Equilibria. Proceedings of the Eleventh International Conference on Artificial Intelligence and Statistics, in PMLR 2:347-354

Related Material