基于EM的朴素贝叶斯分类算法  

Naive Bayesian Classification Based on EM Algorithm

作　　者：张亚萍[1] 陈得宝[1] 侯俊钦[1] 

机构地区：[1]淮北师范大学物理与电子信息学院,安徽淮北235000

出　　处：《宿州学院学报》2010年第11期12-13,23,共3页Journal of Suzhou University

基　　金：安徽省高等学校省级优秀青年人才基金项目(2009SQRZ090);安徽省自然科学基金(090412070)

摘　　要：将EM算法引入到朴素贝叶斯分类研究中,提出一种基于EM的朴素贝叶斯分类算法。首先用未缺失的数据属性的算术均数作为初始值,求得极大似然估计;其次迭代执行算法的E步和M步直至收敛,然后完成缺失数据的填补;最后根据朴素贝叶斯分类算法对数据进行分类。实验结果表明,与朴素贝叶斯分类算法相比,基于EM的朴素贝叶斯分类算法具有较高的分类准确率。A Naive Bayesian classification based on EM algorithm is researched.EM algorithm was used to calculate the principal based on incomplete data,maximum likelihood estimation.Each iterative algorithm includes two steps: the first step in seeking expectations(Expectation Step),known as the E step;the second step for maxima(Maximization Step),known as step-by-step M,then the absent value of the record is filled by the corresponding attribute of the data.Finally,the handled data set is clustered by naive Bayesian...

关 键 词：朴素贝叶斯分类 先验概率 后验概率 EM算法 缺失数据 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论]