Automatic Construction of Static Evaluation Functions for Computer Game Players
Constructing evaluation functions with high accuracy is one of the critical factors in computer game players. This construction is usually done by hand, and deep knowledge of the game and much time to tune them are needed for the construction. To avoid these difficulties, automatic construction of the functions is useful. In this paper, we propose a new method to generate features for evaluation functions automatically based on game records. Evaluation features are built on simple features based on their frequency and mutual information. As an evaluation, we constructed evaluation functions for mate problems in shogi. The evaluation function automatically generated with several thousand evaluation features showed the accuracy of 74% in classifying positions into mate and non-mate.
Unable to display preview. Download preview PDF.
- 3.Kaneko, T., Yamaguchi, K., Kawai, S.: Automated Identification of Patterns in Evaluation Functions. In: van den Herik, H.J., Iida, H., Heinz, E.A. (eds.) Advances in Computer Games, vol. 10, pp. 279–298. Kluwer Academic Publishers, Dordrecht (2004)Google Scholar
- 6.Gomboc, D., Marsland, T.A., Buro, M.: Evaluation fuction tuning via ordinal correlation. In: van den Herik, Iida, Heinz (eds.) Advances in Computer Games, pp. 1–18. Kluwer Academic Publishers, Dordrecht (2003)Google Scholar
- 7.Fayyad, U.M., Keki, B.: Multi-Interval Discretization of Continuous-Valued Attributes for Classification Learning. In: IJCAI 1993, pp. 1022–1027 (1993)Google Scholar
- 9.Sakuta, M., Iida, H.: And/or-tree search algorithms in shogi mating search. ICGA Journal 24(4), 218–229 (2001)Google Scholar