Macroscopic understanding of the game situations in GO

Abstract

Chess program has a great progress by a method of game tree search. In the contrary, the search method does not work well in GO, because GO has much vaster search space. In this paper, we propose that a fuzzy approach can be applied to GO algorithm; especially for understanding of "Atsumi(thickness)". "Atsumi" is treated as mixed fuzzy values of various axes similar to ones human players use when they decide the strategy. We also propose that general rules of GO strategy can be written by "Atsumi" and show that the approach is applicable to open game or beginning of middle game of GO.<>