Description
This book is the tenth in a well-established series, originally describing the progress of computer-chess research only. While chess has dominated AI work in intelligent game-playing for almost half a century, games presumably harder than chess, such as Go, have moved into the spotlight in recent years. The research reported on herein reflects this growing trend with just 6 out of the 24 works overall still focussing on chess, equally many concentrating on Go, and the remaining 12 relating to checkers, Lines of Action, and a variety of other games. Chess – automatic tuning of evaluation functions; modelling of, pattern recognition in, and strategies for endgame play; search and knowledge in endgames; selective pruning of the search; Go – evaluation by neural networks; incremental static analysis; machine learning for position scoring; Monte-Carlo experiments; special applications of DF-PN search; sufficiency of single eyes; Checkers and Lines of Action – endgame databases for checkers; evaluation, search, and knowledge in Lines of Action; Others – an evaluation function for Amazons; opponent-model search in Bao; searching with dependency analysis in Gaps; solving 7??7 Hex; programming of Kriegspiel endings; solving Oshi-Zumo; automatic pattern identification for Othello; new sequences and results for 2-pile subtraction games extending Wythoff’s game. Advances in Computer Games: Many Games, Many Challenges comprises the proceedings of the 10th Advances in Computer Games Conference (ACG-10) that was generously sponsored by the European Union (EU), endorsed by the Specialist Group SG-16 of the International Federation for Information Processing (IFIP), organized incooperation with Kurt Jungwirth from the University of Graz, and held in Graz, Styria, Austria, from November 24-27, 2003. Simultaneously, the 11th World Computer-Chess Championship and the 8th Computer Olympiad took place in Graz as well from November 22-30, 2003.
