Abstract: Chess has always been an attraction for mankind for it’s apparently simple rules but difficult decision making required at different stages of the game.Psychologists have extensively studied chess to look into human cognitive processes and perceptions of the board associated with the game.In the last part of previous century AI scientists were very eager to build a chess engine that can exibit typical human-like chess intelligence. They were successful performance-wise but the human-like perception or decision-making were not at all present in those engines.Realising the non-triviality of the problem,Alexander Kronrod, a Russian AI researcher, said ’Chess is the Drosophila of AI.’Some of them tried to build engines that depends on learning but the then tools for learning approach were limited.After the explosion of machine learning research in last few decades that gave birth to algorithms like SVM,chess AI was revisited.In this paper a machine learning approach towards king-pawn endgames is presented.
Links: