Adaptation of fuzzy rule-based systems for game playing

This paper discusses the adaptation of fuzzy rule-based systems during the iterative execution of a multi-player non-cooperative repeated game. We first briefly describe a market selection game, which is a non-cooperative repeated game with many players and several alternative actions. We also describe some simple strategies. Next we show how our market selection game can be handled as a pattern classification problem where a single training pattern is successively generated from every round of our game. A fuzzy rule-based classification system is used by each player for choosing an action in every round. An incremental learning algorithm is proposed. Then we show how our market selection game can be handled as a function approximation problem. A fuzzy rule-based approximation system, for which an incremental learning algorithm is also proposed, is used as a value function for approximating the expected payoff from each action. Finally, we demonstrate that our two approaches can quickly adapt to a change of the environment in our game.