摘要
UCT(Upper Confidence Bound Apply to Tree)算法,即上限置信区间算法,是对蒙特卡洛算法利用UCB1算法进行改良的随机模拟算法。但对于苏拉卡尔塔棋,有大量的棋子数较多的棋局,单次模拟的随机性很难快速完成一次有效对局。甚至过多随机模拟会使单局的最终结果与真实情况大相径庭。对此利用限制单局模拟次数搭配估值函数来改进UCT的模拟函数,有效的提升了UCT算法的模拟速度和模拟准确性,提高了UCT算法的博弈能力。
UCT(Upper Confidence Bound Apply to Tree)algorithm is a stochastic simulation algorithm modified by Monte Carlo algorithm using UCB1 algorithm.However,in the case of Surakarta,there are a large number of chess games with a large number of pieces,and the randomness of a single simulation is difficult to quickly complete an effective game.Even too many random simulations can make the final result of a single round very different from the real situation.In this regard,the simulation function of UCT is improved by limiting the number of simulations in a single game and the estimation function,which effectively improves the simulation speed and accuracy of the UCT algorithm and improves the game ability of the UCT algorithm.
作者
杨鑫朋
王静文
YANG Xinpeng;WANG Jingwen(School of science,Shenyang University of Technology,Shenyang 110870,China)
出处
《智能计算机与应用》
2025年第12期83-87,共5页
Intelligent Computer and Applications
