USING THE LOOPY BELIEF PROPAGATION IN SIGUO

Article type: Research Article

Authors: Xia, Zhengyou | Zhu, Yongping | Lu, Hui¹

Affiliations: Nanjing, China

Note: [1] Department of computer science, Nanjing University of Aeronautics and Astronautics, 210016, China. Email: zhengyou_xia@yahoo.com. This note is supported by the JiangSu Province Science Technology Foundation under Grant No. BK2006567.

Abstract: Siguo is an oriental game with imperfect information. In this note, we suggest using probability in a Bayesian sense to model the uncertainty regarding the opponent’s types of pieces. We use loopy belief propagation for rapid inference and prediction about the piece types of an opponent. Our experimental results indicate that the method can successfully infer the exact type of the opponent’s piece. The ratio of true inference is approximately 48.7%, which is more than elementary players’ inference strength.

DOI: 10.3233/ICG-2007-30404

Journal: ICGA Journal, vol. 30, no. 4, pp. 209-220, 2007