基于广义鲁棒距离度量的孪生支持向量机分类算法
TWIN SUPPORT VECTOR MACHINE CLASSIFICATION BASED ON GENRALIZED ROBUST DISTANCE METRIC
李耀波 1宋旭东 2孔翔宇3
作者信息
- 1. 辽宁对外经贸学院信息管理学院 辽宁大连 116052
- 2. 大连交通大学软件学院 辽宁大连 116028
- 3. 辽宁对外经贸学院科研处 辽宁大连 116052
- 折叠
摘要
针对孪生支持向量机处理含离群点效果较差的问题,提出一种基于Laplacian核相关熵的广义鲁棒距离度量分类算法.提出一种有界自适应Lθε损失,在学习过程中可以通过自适应参数θ来选择不同的损失函数;提出一种基于拉普拉斯核的相关熵诱导鲁棒距离度量,并证明该度量的有界性、非凸性、光滑性和逼近性;进一步提出一种自适应鲁棒孪生支持向量机学习算法.在多个数据集上的实验结果表明,该算法对特征噪声和离群点具有良好的鲁棒性.
Abstract
Aimed at the problem that twin support vector machine(TWSVM)has poor performance in dealing with outlier classification,a generalized robust distance measure classification method based on Laplacian kernel correlation entropy is proposed.A bounded adaptive Lθε loss function was proposed.In the learning process,different loss functions could be selected by adaptive parameters θ.A correlation entropy induced robust distance measure based on Laplace kernel was proposed,and its boundedness,non-convexity,smoothness and approximation were proved.An adaptive robust twin support vector machine learning framework was introduced.Experimental results on several data sets show that the proposed method is robust to feature noise and outliers.
关键词
支持向量机/有界性/相关熵/鲁棒性Key words
Support vector machine/Boundedness/Correlation entropy/Robustness引用本文复制引用
基金项目
辽宁省自然科学基金项目(2019ZD0105)
出版年
2024
国家科技期刊平台
NSTL
维普
万方数据