| 2,606 | 198 | 21 |
| 下载次数 | 被引频次 | 阅读次数 |
研究了贝叶斯分类器家族中具有代表性的分类器,即朴素(naive)贝叶斯分类器、贝叶斯网络分类器和TAN(treeaugmentednaiveBayesian)分类器;发现属性变量之间的依赖相对于属性变量与类变量之间的依赖是可以忽略的,因此在所有树形分类器中TAN分类器是最优的.
Abstract：The classification is an important and basic ability for human obtained by learning.It has been considered as a key research area in machine learning,pattern recognition and data mining.In this paper,the representative classifiers in Bayes classifier family are studyed.They are naive Bayesian classifiers,Bayesian network classifiers and tree augmented naive Bayesian classifiers .They are also important members in whole classifier family.It is proved that a TAN classifier is optimal in tree classifiers.
[1] 王辉 用于预测的贝叶斯网络[J].东北师大学报(自然科学版),2002,34(1):9～14.
[2] 王辉.用于决策支持的贝叶斯网络[J].东北师大学报(自然科学版),2001,33(4):26～30.
[3] DavidMaxwell.LearningequivalenceclassesofBayesian-networkstructures[J].MachineLearning,2002(2):445～498.
[4] NirFriedman.Bayesiannetworkclassifiers[J].MachineLearning,1997,29:131～163.
[5] MarcoRamoni.RobustBayesclasifiers[J].ArtificialIntelligence,2001,125(1,2):209～226.
[6] DavidHeckerman.LearningBayesiannetworks:thecombinationofknowledgeandstatisticaldata[J].MachineLearning,1995,20:197～243.
[7] ChengJie.LearningBayesiannetworksfromdata:aninformation-theorybasedapproach[J].ArtificialIntelligence,2002,137(1,2):43～90.
[8] WongML.UsingevolutionaryprogrammingandmininumdescriptionlengthprinciplefordataminingofBayesiannetworks[J].IEEETransactionsonPatternAnalysisandMachineIntelligence,1999,21(2):174～178.
基本信息:
中图分类号:TP183
引用信息:
[1]周颜军,王双成,王辉.基于贝叶斯网络的分类器研究[J].东北师大学报(自然科学版),2003(02):21-27.
基金信息:
科学技术部国家软科学研究项目(Z99015)
2003-06-23
2003-06-23