一种基于强属性限定的贝叶斯分类模型

A Restricted Bayesian Classification Model Based on Strong Attributes

朴素贝叶斯分类模型一种简单而高效的分类模型,但它的条件独立性假设使其无法将属性间的依赖表达出来,影响了它分类的正确率。属性间的依赖关系与属性本身的特性有关,有些属性的特性决定了其他属性必然依赖于它,即强属性。文中通过分析属性相关性的度量和贝叶斯定理的变形公式,介绍了强属性的选择方法,通过在强弱属性之间添加增强弧以弱化朴素贝叶斯的独立性假设,扩展了朴素贝叶斯分类模型的结构。在此基础上提出一种基于强属性限定的贝叶斯分类模型SANBC。实验结果表明,与朴素贝叶斯分类模型相比,SANBC分类模型具有较高的分类正确率。

Naive Bayesian classification model is a simple and effective classification model, but its attribute independence assumption makes it unable to express the dependence among attributes, and affects its classification accuracy. The inter- dependence between attributes is closely related to their features, i.e. the features of some entail the others＇ dependence upon them - strong attributes. The paper presents SANBC（A Restricted Bayesian Classification Mmodel Based on Strong Attributes） following the extension of structure of naive Bayesian classification model, through the analysis of a variant of Bayes theorem, the evaluation of condition attribute with correlation, and the instruction of the selection of strong attributes and the attribute independence assumption that naive Bayesian classification model can be weakened through the adding of highlighting lines between strong and weak attribute. Compared with Bayesian classification model, experimental results show SANI3C has higher accuracy.