摘要
计算机博弈是人工智能领域的挑战性课题,它利用计算机进行分析、判断和推理,从而得到理性的决策.不围棋是近年来计算机博弈竞赛的一个棋种,属于围棋的变体,其规则是先吃子或棋子自杀的一方为负.通过分析不围棋博弈模型的特点,提出了对上限信心界树搜索(UCT)算法的一个优化方法,在算法的启动过程优先选择评分较高的盘面进行模拟博弈,以便得到更好的落子选择.在与著名的OASE-NoGo软件的试验对弈中,以该算法为核心设计的不围棋软件取得了90%以上的胜率,证明是可行、有效的.
Computer game is a challenging task in the field of artificial intelligence, which makes use of computer to analyze, judge and reason, so as to get the rational decision. In recent years, NoGo game is a kind of computer game competitions, similar to the game of Go, with the rules of no capture or suicide which is allowed. According to the characteristics of NoGo game model, an optimized UCT(UCB for Tree Search) algorithm is proposed to compute the better move by choosing certain chess layout with high score preferentially. A NoGo game software is designed mainly relies on the above-mentioned algorithm and proved to be feasible and efficient by the result of achieving more than 90% of winning with the famous OASE-NoGo.
出处
《韶关学院学报》
2015年第8期17-21,共5页
Journal of Shaoguan University
基金
国家级大学生创新创业训练计划项目(201410576018)
广东省大学生创新创业训练计划项目(201410576060)
韶关学院科研项目(2012-16)