Fuzzy extraction of human features represented in the game record

Abstract

For two person zero sum perfect information games such as chess and shogi, game AI that exceeds the top players has been developed. On the other hand, it is difficult to understand the move intentions and aims of a strong game AI, and sometimes humans feel discomfort when playing against it. Therefore, as a preliminary step toward realizing a chess AI that plays moves understandable to humans, this study examines the characteristics of humans by using the game records of both humans and AI. In particular, we compare human and AI game records, focusing on the human emotion of impatience, and perform fuzzy extraction on the found features to show impatience in a form understandable to humans.