基于时间序列的竞技水平估计算法被引量：1

A skill estimation algorithm based on time series

下载PDF

导出

摘要体育竞技中,选手水平随时间变化,True Skill-T算法能估计选手时间序列上的竞技水平。文中针对围棋比赛的特点对其进行简化和修改,实现了True Skill Go-Year算法。为研究算法在实际排名应用上的可行性,分别基于真实的围棋比赛数据和仿真数据,考察了算法对数据的适应能力。实验结果表明True Skill-T算法排名结果符合领域常识,具备实际应用的性能。 As player skills changes over time in competitive sports, TrueSkill -T has be suggested to infer entire time series of player skills. According to the characteristics of Go matches, TrueSkillGo-Year model was implemented by simplifying and modifying TrueSkill - T. The adaptability of this algorithm to data was investigated for practical applications. The data sets used in the experiments are the real world and simulated data of Go games. The experimental results show that ranking processed by TrueSkill - T is consistent with domain knowledge and can be used in practical applications.

作者陈磊吴霖袁梅宇江虹

机构地区昆明理工大学信息工程与自动化学院计算机系

出处《信息技术》 2015年第5期32-37,共6页 Information Technology

基金国家自然科学基金(61163004) 云南省应用基础资助项目(2010CD027)

关键词竞技水平估计时间序列概率图模型 TRUE Skill-T skill estimation time series graphical model TrueSkill - T

分类号 TP301.6 [自动化与计算机技术—计算机系统结构]

引文网络
相关文献

参考文献10

1Elo A E. The Rating of Chess Players, Past and Present [ M ]. New York : Arco Publishing, 1978.
2Clickman M E. Parameter estimation in large dynamic paired com- parison experiments [ J ]. Applied Statistics, 1999,48:377 - 394.
3Herbrich, R, Minka, T, Graepel T TrueskillTM. A Bayesian skill rating system [ C ]//In Advances in Neural Information Processing Systems. 2007:569 -576.
4Pierre Dangauthier, Ralf Herbrich, Tom Minka, et al. TrueSkillthrough time: revisiting the history of chess [ C ]//Advances in Neu- ral Information Processing Systems 21,2008:337 -344.
5Kschisehang F R, Frey B J, Loeliger H A. Factor graphs and the sum-product algorithm [ J ]. IEEE Transactions on, Information Theory, 2001,47(2) :498 -519.
6Minka, Thomas P. Expectation propagation for approximate Baye- sian inferonce [ C ]. Proceedings of the Seventeenth conference on Uncertainty in artificial intelligence. Morgan Kaufmann Publishers Inc., 2001:362-369.
7吴霖,陈磊,邓超,袁梅宇,江虹.基于TrueSkill模型的围棋棋手排名方法及评估[J].昆明理工大学学报（自然科学版）,2013,38(3):47-55. 被引量：5
8NihonKiin.棋士[EB/OL].[2014-03-12].http://www.ni-honkiin.or.jp/player/index.html.
9Taiwango.棋士资讯[EB/OL].[2014-03-12].http://WWW.taiwango.org.tw/chesser.asp.
10Tom.资料汇编[EB/OL].[2014-03-12].http://weiqi.com.com/secondary/wqzl/wqzl-index.htm.

二级参考文献13

1中国围棋竞赛规则(2002年版)[EB/OL].[2012—10—16].http://games.sports.cn/datebase/encyelopaedia/wq/2010—07—19/2046610_3.html.
2Taeil Bai. The New Rating System of the Korea Baduk Association (EB/OL).[2012 - 11 - 12]. http://english. baduk. or. kr/sub03_04 -1. htm? menu = f13&divL =6.
3Herbrich R, Minka T, Graepel T. TrueSkill (TM) : A Bayesian skill rating system[C] I lIn Advances in Neural Information Pro?cessing Systems 20. Scholkopf B, PlattJ, Hoffman T. Cambridge: MIT Press, 2007 :569 - 576.
4Zhang Xinhua, Graepel T, Ralf Herbrich. Bayesian Online Learning for Multi - label and Multi - variate Performance Meas?ures[C] II Proceedings of the 13 th International Conference on Artificial Intelligence and Statistics (AIST ATS) , 2010. Yee Whye The, Mike Titterington. Brookline: Microtome Publishing, 2010:956 -963.
5Graepel T, CandelaJ Q, Borchert T, Herbrich R. Web - scale Bayesian click - through rate prediction for sponsored search advertising in microsoft's bing search engine[C] IIProceedings of the 27th International Conference on Machine Learning IC?ML 2010.Johannes Furnkranz , ThorstenJoachims. Anderson: Omnipress , 2010: 13 - 20.
6RANKINGS (EB/OL).[2012 -10 -15]. http://www.atpworldtour.comiRankings/Rankings - FAQ. aspx.
7Elo A E. The new U. S. C. F. rating system[J]. Chess Life, 1961,16: 160 -161.
8Elo A E. The Rating of Chess Players, Past and Present[M]. New York: Areo Publishing, 1978.
9Clickman M E. A Comprehensive guide to chess ratings[J]. Amer. ChessJournal, 1995 (3) :59 - 102.
10Minka T P. Expectation propagation for approximate Bayesian inference[CJ II Proceedings of the Seventeenth conference on Uncertainty in artificial intelligence.John Breese, Daphne Koller. San Francisco: Morgan Kaufmann Publishers Inc. , 2001 : 362 -369.