Reinforcement Learning to Play an Optimal Nash Equilibrium in Team Markov Games Export

@inproceedings{citeulike:5845570, abstract = {Multiagent learning is a key problem in game theory and AI. It involves two interrelated learning problems: identifying the game and learning to play. These two problems prevail even in team games where the agents\&\#039; interests do not conflict. Even team games can have multiple Nash equilibria, only some of which are optimal. We present optimal adaptive learning (OAL), the first algorithm that converges to an optimal Nash equilibrium for any team Markov game. We provide a convergence proof, and show that the algorithm\&\#039;s parameters are easy to set so that the convergence conditions are met. Our experiments show that existing algorithms do not converge in many of these problems while OAL does. We also demonstrate the importance of the fundamental ideas behind OAL: incomplete history sampling and biased action selection.}, author = {Wang, Xiaofeng and Sandholm, Tuomas}, booktitle = {in Advances in Neural Information Processing Systems}, citeulike-article-id = {5845570}, citeulike-linkout-0 = {http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.19.2669}, keywords = {2002, convergence, cooperation, game-theory, markov, oal, reinforcement-learning}, pages = {1571--1578}, posted-at = {2009-09-27 17:33:05}, priority = {0}, title = {Reinforcement Learning to Play an Optimal Nash Equilibrium in Team Markov Games}, url = {http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.19.2669}, volume = {15}, year = {2002} }

TY - CONF ID - citeulike:5845570 L3 - citeulike-article-id:5845570 N2 - Multiagent learning is a key problem in game theory and AI. It involves two interrelated learning problems: identifying the game and learning to play. These two problems prevail even in team games where the agents' interests do not conflict. Even team games can have multiple Nash equilibria, only some of which are optimal. We present optimal adaptive learning (OAL), the first algorithm that converges to an optimal Nash equilibrium for any team Markov game. We provide a convergence proof, and show that the algorithm's parameters are easy to set so that the convergence conditions are met. Our experiments show that existing algorithms do not converge in many of these problems while OAL does. We also demonstrate the importance of the fundamental ideas behind OAL: incomplete history sampling and biased action selection. EP - 1578 TI - Reinforcement Learning to Play an Optimal Nash Equilibrium in Team Markov Games VL - 15 T2 - in Advances in Neural Information Processing Systems SP - 1571 KW - 2002 KW - convergence KW - cooperation KW - game-theory KW - markov KW - oal KW - reinforcement-learning AU - Wang, Xiaofeng AU - Sandholm, Tuomas PY - 2002/// UR - http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.19.2669 ER -