[1]岳 鹏,李太华,邱玉辉.对称无关的模式编码方法在定式学习中的应用[J].智能系统学报,2007,2(04):92-94.
 YUE Peng,LI Tai hua,QIU Yu hui.A pattern encoding method independent of symmetry applied in joseki learnin g[J].CAAI Transactions on Intelligent Systems,2007,2(04):92-94.
点击复制

对称无关的模式编码方法在定式学习中的应用(/HTML)
分享到:

《智能系统学报》[ISSN:1673-4785/CN:23-1538/TP]

卷:
第2卷
期数:
2007年04期
页码:
92-94
栏目:
出版日期:
2007-08-25

文章信息/Info

Title:
A pattern encoding method independent of symmetry applied in joseki learnin g
文章编号:
1673-4785(2007)04-0092-03
作者:
岳  鹏李太华邱玉辉
西南大学智能软件与软件工程重点实验室,重庆400715
Author(s):
YUE PengLI TaihuaQIU Yuhui
The Key Laboratory of Intelligent Software & Software Engineering, Southwest Un iversity, Chongqing 400715, China
关键词:
计算机围棋定式
Keywords:
computer Go joseki
分类号:
TP181
文献标志码:
A
摘要:
计算机围棋中的定式可以从棋谱中学习,采用与模式对称性无关的编码方法可以取得更好的储存效果.这是一种基于模式的邻近特征、轮廓特征和行列特征的模式编码方法,与模式的平移、旋转、翻转及黑白对称性无关.实验结果表明其哈希性能良好,模式间的查找迅速,也可以在官子和死活问题中得以应用.
Abstract:
Joseki can be learned from game records in Computer Go. To more effect ively maintain the joseki library, a pattern encoding method independent of symm etry is needed. This is provided by an encoding method is designed based on the pattern’s neighbor character, the outline character and the row and column’s c haracter. The results show that this method’s hashing performance is good, and it rapidly switches between patterns. The method can also be applied to tsumego and yose in Computer Go. 

参考文献/References:

[1] ALBERT L.Feature extractions and representation for pattern recog ni tion and the game of Go[D]. Graduate School of the University of Wisconsin, 19 70.
[2]谷 蓉,刘学民,朱仲涛,周 杰.一种围棋定式的机器学习方法[J].计算机工程,2004,30(6):142-144.
 GU Rong, LIU Xuemin, ZHU Zhongtao, ZHOU Jie. A machine learning method of josek i in Go[J]. The Journal of Computer Engineering, 2004, 30(6): 142-144.
[3]GRAEPEL T. PACBayesian pattern classification with kernels : theory, algorithms, and an application to the game of Go[D].Berlin: Technical University of Berlin, Berlin, 2002
[4]BALDI P, RALAIVOLA L, WU L. SVM and patternenriched common fate graphs fo r the game of Go[A]. European Symposium on Artificial Neural Networks[C].Brug es, B elgium, 2005.
[5]STOUTAMIRE D. Machine learning, game play, and Go[D]. Case Weste rn Reserve University, 1991

相似文献/References:

[1]张培刚,陈克训.使用不同的博弈树搜索算法解决计算机围棋的吃子问题[J].智能系统学报,2007,2(03):84.
 ZHANG,Pei-gang,CHEN Keh-hsun.Using Different Search Algorithms to Solve Computer Go Capturing Problems[J].CAAI Transactions on Intelligent Systems,2007,2(04):84.

备注/Memo

备注/Memo:
收稿日期:20060821.
作者简介:
岳 鹏,男,1974年生,博士研究生,主要研究方向为博弈,已发表论文2篇. E-mail: yappy555@swu.edu.cn.
 李太华,男,1974年生,博士研究生,主要研究方向为多Agent系统,已发表论文2 篇.E-mail: catalyst@swu.edu.cn.
邱玉辉,男,1938年生,博士生导师,中国人工智能学会副会长,主要研究方向为模糊逻辑、多Agent系统、博弈等,主持主研国家自然科学基金、省市自然科学基金及攻关项目15项,曾荣获曾宪梓教育基金会三等奖,四川省优秀教学成果一、二等奖,重庆市优秀教学成果一等奖,重庆市自然科学奖二等奖,重庆市科技进步二、三等奖,1992年起享受国家特殊津贴.已发表论文180余篇,出版学术专著10余部. E-mail:yhqiu@swu.edu.cn.
更新日期/Last Update: 2009-05-07