SELF-PLAYING-BASED OPENING BOOK TUNING
Jun Nagashima, Tsuyoshi Hashimoto, Hiroyuki Iida · New Mathematics and Natural Computation · 2006
This paper proposes a method to tune an opening book based on self-playing games. In this method, many self-playing games are performed from various positions that are contained in the book. The book is well tuned for the target computer program that it may lure the opponent to play into its own hands. We produced some different types of books through the self-playing based tuning to perform a tournament. The tournament results indicate a remarkable playing strength improvement compared to the program with the original book. The proposed idea was implemented in our computer shogi program TACOS, which won the latest Computer Olympiad.