基于信息增益的最优组合因子Fisher判别法  

Optimal combination of factor Fisher discrimination method based on information gain

作　　者：毛临川 吴根秀[1] 吴恒[1] 黄梅[1] MAO Linchuan;WU Genxiu;WU Heng;HUANG Mei(School of Mathematics and Information Science, Jiangxi Normal University, Nanchang 330022, China)

机构地区：[1]江西师范大学数学与信息科学学院,南昌330022

出　　处：《计算机工程与应用》2016年第19期94-96,145,共4页Computer Engineering and Applications

基　　金：江西省教育厅科学技术项目(No.GJJ14244)

摘　　要：线性判别分类器是一种有效的模式分析技术,其中以Fisher判别法准则应用最广,目前已有多种改进线性提取方法。引进信息增益,建立基于信息增益的最优组合因子判别分类器,实现最优组合因子判别分类器的优化。实验结果表明优化后的分类器有效剔除了冗余因子,具有良好的分类准确率。Linear discriminant classifier is an effective model analysis technology, among them with Fisher discriminantcriterion, the most widely used, there are many kinds of improved linear extraction method. This paper introduces the informationgain, to establish the optimal combination of the factors based on the information gain discriminant classifier,quickly chooses the optimal discriminant classifier combination factor. Experiments show that the rapid selected classifiereliminates the redundancy factor effectively, and has good classification accuracy.

关 键 词：信息增益 最优组合因子 回代率 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]