One of the fundamental mathematical constructs is the game. Formally, two or more players (of which one may be "nature") have resources to allocate and receive a payoff for their allocations. Each player may desire to maximize his or her immediate payoff, long-term payoff, or may be concerned with more complex issues that relate to other players (e.g., maximize collective payoff, ensure minimizing payoff to an opponent). Evolutionary computation has proven to be an interesting and effective tool in machine learning methods to address games of many forms. These include the iterated prisoner's dilemma, standard board games, military simulations, and other instances. Evolutionary algorithms have been used to learn effective strategies against both fixed and simultaneously evolving opponents (co-evolution), in cases of complete and also incomplete, uncertain, and noisy information about the environment of the game. Many open issues have been identified, including but not limited to the selection of evolvable representations, choosing opponents effectively to promote evolutionary learning, and the requirements sustained co-evolutionary arms races and open-ended evolution. The special issue will entertain submissions in all areas of evolutionary computation and games. Substantially extended and revised conference papers will be considered.
Deadline for Submission: July 31, 2004
Final Decisions: Jan. 31, 2005
Special Issue appears Summer, 2005