【摘要】 朴素贝叶斯分类的缺点在于过于强调属性间的独立性和属性对分类影响的一致性,无法真实表达属性间的相关性,影响了这种方法的健壮性和分类效果。信息增益是属性对分类重要性的一种度量方法,一方面通过对属性互斥信息的分析,去除与分类无关的属性,简化朴素贝叶斯分类器的结构;另一方面通过对属性信息增益的度量,选择对分类有较大影响的关键属性,在关键属性与非关键属性之间通过添加有向边的方式扩展朴素贝叶斯分类器的结构,放宽属性间独立性的限制,提高分类性能。更多还原

【Abstract】 The disadvantage of naive bayes classification is that too much emphasis is placed on the independence of attributes and the consistency of attributes’ influence on classification,and the correlation between attributes can not be truly expressed,which affects the robustness and classification effect of this method. Information gain is a measure of the importance of attributes to classification. On the one hand,by analyzing the mutually exclusive information of attributes,the attributes unrelated to classification are removed and the structure of Naive Bayes classifier is simplified On the other hand,by measuring the gain of attribute information,we can select the key attributes which have great influence on classification,and extend the structure of the Naive Bayes classifier by adding oriented edges between the key attributes and the non-key attributes,the restriction of independence between attributes is relaxed to improve the classification performance.更多还原