摘要
In two player games like professional chess and shogi in the real world, the players play against each other repeatedly while changing their strategies and striving for mastery. The relation between these players can be ascribed to competitive coevolution represented by predator-prey or parasite-host. In this paper, we propose conditions for players to realize the reciprocal development of their strategies in two player games. We construct a co-evolutionary system that reciprocally develops players' strategies. The game environment is the seven card stud poker, which is one of the complex real world games of imperfect information. In our system, the players decide their actions based on self-learning by Classifier Systems and then make the strategies more sophisticated. We analyze dynamics of the evolution of the player's strategies and show the learning process of reciprocating skills of players.
- 出版日期2008-12