基于逆云模型的支持向量机多类分类方法  

SVM multi-class classification based on backward cloud model

作　　者：刘婷[1] 陈晓云[1] 

机构地区：[1]福州大学数学与计算机科学学院,福建福州350002

出　　处：《福州大学学报（自然科学版）》2008年第3期341-346,共6页Journal of Fuzhou University(Natural Science Edition)

基　　金：福建省自然科学基金资助项目(S0650013)

摘　　要：针对支持向量机在进行多类识别中存在的拒分问题,结合能够描述客观世界不确定性的模糊理论,提出利用逆云模型描述样本对各类逆云隶属度的多类分类方法.实验表明,该方法能够有效地消除样本拒分现象,提高识别率.Support vector machine is a new pattern recognition method based on statistical learning theory. For the existence of rejected classification problem in the multi - classification , and combining with the fuzzy theory which can describe the unsureness of object world, a multi - class classification method using backward cloud model to compute backward cloud membership between swatches and every class is proposed in this paper. Simulations demonstrate that this approach can avoid rejected classification swatches effectively and improve recognition rate consequently.

关 键 词：支持向量机 多类识别 逆云模型 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]