Opponent Provocation and Behavior Classification: A Machine Learning Approach

摘要: Opponent Modeling is one of the most attractive and practical arenas in Multi Agent System (MAS) for predicting identifying future behaviors opponent. This paper introduces a novel approach using rule based expert system towards opponent modeling RoboCup Soccer Coach Simulation. In this scene, an autonomous coach agent able to identify patterns by analyzing opponent's past games advising own players. For purpose, main goal our research comprises two complementary parts: (a) developing 3-tier learning architecture classifying behaviors. To achieve objective, sequential events game are identified environmental data. Then predicted statistical calculations. Eventually, comparing with rest team's behavior, model constructed. (b) designing containing provocation strategies expedite detection patterns. These items mentioned used coach, generate appropriate strategy play against structure tested Simulation MRLCoach was champion at 2006 Germany.