一种具有特征有效度的模糊支持向量机  

Fuzzy Support Vector Machine with Feature Weighted Degree

作　　者：顾成杰[1] 张顺颐[1] 黄河[2] 孙雁飞[1] 

机构地区：[1]南京邮电大学信息网络技术研究所,江苏南京210003 [2]北京航空航天大学软件学院,北京100083

出　　处：《南京邮电大学学报（自然科学版）》2011年第3期50-54,61,共6页Journal of Nanjing University of Posts and Telecommunications：Natural Science Edition

基　　金：国家高技术研究发展计划(863计划)(2006AA01Z232;2009AA01Z212;2009AA01Z202);国家自然科学基金(61003237);江苏省自然科学基金(BK2007603);江苏省高技术研究计划(BG2007045);江苏省重大科技支撑计划(BE2008134);江苏省科技成果转化专项基金(BA2007012)资助项目

摘　　要：针对传统支持向量机对噪声或野值样本敏感和忽略样本各个特征对分类精度的不同影响,提出了一种具有特征有效度的模糊支持向量机(FW-FSVM)。该方法通过对模糊支持向量机的模糊因子进行改进,将噪声或野值样本与有效样本进行区分。在此基础上根据信息增益计算样本各个特征的特征有效度,消除弱相关特征和冗余特征对分类精度的影响,从而构建了具有特征有效度的模糊支持向量机。与以往的SVM改进方法相比较,实验验证了FW-FSVM的有效性,表明其具有更好的分类精度和泛化能力。Since the traditional support vector machines are sensitive to noises or outliers and neglect the relative importance of each feature with classification accuracy,a novel fuzzy support vector machine with feature weighted degree（FW-FSVM） is proposed in this paper.This approach improves the feature factor on the basis of fuzzy support vector machine to classify the noise samples and outlier samples from the efficient samples.In order to diminish the effect of weak features or redundancies for classification accuracy,the proposed approach calculates the feature weighted degree of each feature by computing the information gain.A fuzzy support vector machine with feature weighted degree can be constructed.Compared with the traditional variants of SVM,FW-FSVM is validated by experimental results.It can improve the classification accuracy and generalization performance greatly.

关 键 词：模糊支持向量机 模糊因子 特征有效度 分类精度 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]