摘要
孪生支持向量机通过求解较小的二次规划问题,提高了分类器的性能,然而,该方法主要利用了类间可分的特性,并使用hinge损失函数构建相应的模型,它们并未充分考虑不同类中数据的结构信息以及不同样本对分类的影响,导致该方法对噪声具有较强的敏感性以及重取样的不稳定性.为了进一步提高孪生支持向量机的性能,基于pinball损失函数,将数据集中不同类的结构信息以及不同样本的作用引入到孪生支持向量机中,获得了基于pinball损失的结构模糊孪生支持向量机模型,从理论上导出了基于pinball损失的结构模糊孪生支持向量机算法pin-sftsvm,通过选取人工生成数据集与UCI标准数据集,对pin-sftsvm算法进行了实验,并与tbsvm、s-tsvm和pin-tsvm算法进行了性能比较,表明了提出算法的有效性.
Twin support vector machine improves the performance of the classifier by solving the smaller quadratic programming problem.However,this method mainly utilizes the separability between classes and constructs the corresponding model using the hinge loss function.Not considering the structural information of the intra-class data and the influence of different samples on the classification,the method has strong sensitivity to noise and instability of resampling.In order to further improve the performance of the twin support vector machine,the structural information of different classes in the data and the effects of different samples are introduced into the twin support vector machine based on the pinball loss function,and the structure fuzzy support vector machine model based on pinball loss is obtained.The structural fuzzy twin support vector machine algorithm pin-sftsvm based on the pinball loss is derived theoretically.The presented algorithm pin-sftsvm is tested by selecting the artificially generated data set and the UCI standard data set,and compared with the tbsvm,s-tsvm and pin-tsvm algorithms.Experimental results show the effectiveness of the proposed algorithm.
作者
李凯
李慧
LI Kai;LI Hui(School of Cyber Security and Computer,Hebei University, Banding, Hebei 071000, China)
出处
《电子学报》
EI
CAS
CSCD
北大核心
2019年第10期2221-2227,共7页
Acta Electronica Sinica
基金
河北省自然科学基金(No.F2018201060)
