基于MWST+T-K2结构学习算法的贝叶斯分类器  

Based on MWST+T-K2 Algorithm to Build Bayes Classifier

作　　者：赵高长[1] 王欣[1,2] 张仲华[1] 韩苗[1] 魏嵬[3] ZHAO Gaochang WANG Xin ZHANG Zhonghua HAN Miao WEI Wei(College of Sciences, Xi＇an University of Science and Technology, Xi＇an 710054, China Yangshengtang CO. ,Ltd. , Hangzhou 310007, China School of Computer Science and Engineering, Xi＇an University of Technology, Xi＇an 710048, China)

机构地区：[1]西安科技大学理学院,西安710054 [2]养生堂有限公司,杭州310007 [3]西安理工大学计算机科学与工程学院,西安710048

出　　处：《复旦学报（自然科学版）》2017年第1期48-56,共9页Journal of Fudan University：Natural Science

基　　金：国家自然科学基金(41271518);陕西省教育厅科研计划项目(2013JK0583;14JK1474);陕西省自然科学基金(2016JM1025)

摘　　要：针对K2算法在构建贝叶斯分类器时节点排序不同影响分类准确率的问题,提出了一种MWST+T-K2结构学习算法,运用Matlab软件的BNT工具箱构建了MWST+T-K2分类器,并经过NBC、TANC、MWST和MWST+T-K2分类器对UCI数据库的24个分类数据集进行分类检验.结果表明,对4种分类器在24个数据集上的分类水平进行整体与两两比较时,MWST+T-K2分类器的分类水平均最优;在小数据集上比较时,MWST+T-K2分类器的分类水平取得全局最优,未取得局部最优;在大数据集上比较时,未取得全局或局部最优,低于TANC的分类水平.所以,MWST+T-K2结构学习算法是一种适合构建小数据集贝叶斯分类器的方法.The node ordering heavily affects the classified accuracy when one constructs Bayes classifier by K2 algorithm.Based on this problem,a kind of MWST＋T-K2 structure learning algorithm is proposed,then this algorithm is applied to construct the MWST＋T-K2 classifier through the BNT toolbox of MATLAB.At last,the classified test is carried out on the 24 data sets in the UCI database.The results imply that the classified level of the MWST＋T-K2 classifier is globally optimal when the four classified levels is compared together or one is compared with anther on the 24 data sets;It is globally（not locally）optimal when the classifier level is compared with the others on all of the small data sets.The result,which is lower than the classified level of TANC,is neither globally nor locally optimal on all of the large data sets.Therefore,the proposed MWST＋T-K2 structure learning algorithm is suitable to construct the Bayes classifier on small data sets.

关 键 词：贝叶斯网络 贝叶斯分类器 MWST+T-K2算法 分类检验 

分 类 号：TP39[自动化与计算机技术—计算机应用技术]