Beyond Samuel: Evolving a Nearly Expert Checkers Player
Evolutionary algorithms can be used to learn how to play complex games of strategy without relying on human expertise. Here I discuss the use of evolutionary computation and artificial neural networks in learning how to play checkers. Starting from neural networks that were created randomly, an evolutionary algorithm has been able to craft a network that can play checkers at a nearly expert level. No features beyond the positions of pieces on the board and the piece differential were provided. The evolutionary algorithm learned everything else on its own, simply by playing the game.
KeywordsHide Layer Evolutionary Algorithm Output Node Sequential Rating Game Tree
Unable to display preview. Download preview PDF.
- Fogel, D.B., Evolutionary Computation: Toward a New Philosophy of Machine Intelligence, 2nd ed., IEEE Press, Piscataway, NJ, 2000.Google Scholar
- Schaeffer, J., One Jump Ahead: Challenging Human Supremacy in Checkers, Springer, Berlin, 1996.Google Scholar
- Chellapilla, K. and Fogel, D.B. “Evolving neural networks to play checkers without relying on expert knowledge,” IEEE Transactions on Neural Networks, 10(6), 1999, in press.Google Scholar
- Fogel, D.B. (ed.), Evolutionary Computation: The Fossil Record, IEEE Press, Piscataway, NJ, 1998.Google Scholar