作者: Gian Piero Favini , Paolo Ciancarini
DOI:
关键词: Extensive-form game 、 Bayesian game 、 Game theory 、 Information set 、 Artificial intelligence 、 Mathematics 、 Non-cooperative game 、 Complete information 、 Perfect information 、 Sequential game
摘要: We describe a novel approach to incomplete information board games, which is based on the concept of metaposition as merging very large set possible game states into single entity contains at least every state in current set. This operation allows an artificial player apply traditional perfect theory tools such Minimax theorem. We this technique Kriegspiel, variant chess characterized by strongly players cannot see their opponent's pieces but can only try guess positions listening messages referee. provide general representation Kriegspiel through trees and weighed maximax algorithm for evaluating metapositions. have tested our competing against both human computer players.