摘要
机器博弈被认为是人工智能领域最具挑战性的研究方向之一。中国象棋计算机博弈的难度绝不亚于国际象棋,但是涉足学者太少,具有自学习能力的就更少了。介绍了中国象棋人机对弈原理,给出了近年来几类典型的评估函数自学习方法及其原理,通过比较得出了最适合中国象棋使用的学习方法。分析了这些方法尚存在的问题,并提出了未来的研究方向。
Computer game is one of the most challenging topics in the field of artificial intelligence. Chinese chess computer game is more complex than chess computer game, and the fewer researchers and the fewest research have the ability of self- teaching in this field. The brief principle of Chinese chess computer game is introduced. The methods of self - teaching of evaluation function in the last few years are presented, as well as get the self- teaching method that fit in with Chinese chess computer game. The general problems with these methods, and promising avenues for future research are discussed.
出处
《计算机技术与发展》
2007年第12期76-79,共4页
Computer Technology and Development
基金
国家自然科学基金(60075019)
关键词
中国象棋
激励学习
神经网络
瞬时差分
博弈
Chinese chess
reinforcement learning
neural network
temporal difference
game
