Playing General Structure Rewriting Games
Łukasz Kaiser, Łukasz Stafiniak · 2010
Achieving goals in a complex environment in which many players interact is a general task demanded from an AI agent.When goals of the players are given explicitly, such setting can be described as a multi-player game with complete information.We introduce a general model of such games in which states are represented by relational structures (hypergraphs), possibly with real-valued labels, and actions by structure rewriting rules.For this model, we develop an algorithm which computes rational strategies for the players.Our algorithm can be parametrized by a probabilistic evaluation function and we devise a general procedure for learning such evaluations.First tests on a few classical examples substantiate the chosen game model and our algorithm.