空间秩深度加权的鲁棒性支撑向量机

Journal of Systems Engineering and Electronics ›› 2009, Vol. 31 ›› Issue (10): 2429-2433,2505.

空间秩深度加权的鲁棒性支撑向量机

胡永刚, 吴翊, 李强

国防科技大学数学与系统科学系, 湖南, 长沙, 410073

收稿日期:2008-09-03 修回日期:2008-11-03 出版日期:2009-10-20 发布日期:2010-01-03
作者简介:胡永刚(1976- ),男,博士研究生,主要研究方向为数据挖掘、遥感图像处理与融合.E-mail:xd7688@hotmail.com
基金资助:
国家自然科学基金(60673090,60802079)资助课题

Robust support vector machine weighted by spatial rank depth

HU Yong-gang, WU Yi, LI Qiang

Dept. of Mathematics and System Science, National Univ. of Defense Technology, Changsha 410073, China

Received:2008-09-03 Revised:2008-11-03 Online:2009-10-20 Published:2010-01-03

摘要/Abstract

摘要： 提出了一种基于数据深度加权的鲁棒性支撑向量机.针对支撑向量机对噪声或离群点的高敏感性,在其优化函数的正则项中增加深度因子,以针对性地弱化这些点对分类结果的影响,使得分类更具有鲁棒性.给出了1范数和2范数约束下的深度加权支撑向量机的具体形式,并推导了空间秩深度在特征空间中的求解方式.相对于文献中的中心距加权方式,该方法抗噪声能力更强.

Abstract: This paper presents a robust support vector machine(SVM) weighted by the spatial rank depth.According to the high sensitivity of SVM to the outliers,the regularization term is weighted by the data depth factor in order to adaptively reduce the influence on the classification by those outliers.The two dual forms of 1-norm and 2-norm depth weighted SVM(DWSVM) are deduced respectively,and the solution of depth in feature space is also presented.Compared with the method weighted by the distance to the mean center,this method is more robust than fuzzy SVM.

中图分类号:

TP391

胡永刚, 吴翊, 李强. 空间秩深度加权的鲁棒性支撑向量机[J]. Journal of Systems Engineering and Electronics, 2009, 31(10): 2429-2433,2505.

HU Yong-gang, WU Yi, LI Qiang. Robust support vector machine weighted by spatial rank depth[J]. Journal of Systems Engineering and Electronics, 2009, 31(10): 2429-2433,2505.

参考文献

[1] Vapnik V N.The nature of statistical learning theory[M].New York:Springer,1995.
[2] Cristianini N,Shawe-Taylor J.An introduction to support vector machines and other kernel-based learning methods[M].West Nyack,NY:Cambridge University Press,2000.
[3] Axelberg P M.Support vector machine for classification of voltage disturbances[J].IEEE Trans.on Power Delivery,2007,22(3):1297-1303.
[4] Zheng S.Multisource image fusion method using support value transform[J].IEEETrans.on Image Processing,2007,6(7):1831-1839.
[5] Burges C.A tutorial on support vector machines for pattern recognition[J].Data Mining and Knowledge Discovery,1998,2(2):1-43.
[6] Lin C F,Wang S D.Fuzzy support vector machines[J].IEEE Trans.on Neural Networks,2002,13(2):464-471.
[7] Zuo Y,Serfling R.General notions of statistical depth function[J].The Annals of Statistics,2000,28(2):461-482.
[8] Zuo Y J,Cui H.Statistical depth functions and some applications[J].Advance in Mathematics(China),2004,33(1):1-25.
[9] Tukey J M.Mathematics and picturing data[C]//Proc.of International Congress on Mathematics,Vancouver,1975,2:523-531.
[10] Wellmann R,Katina S.Calculation of simplicial depth estimators for polynomial regression with applications[J].Computational Statistics & Data Analysis,2007,51(10):5025-5040.
[11] Gao Yonghong.Data depth based on spatial rank[J].Statistics & Probability Letters,2003,65(3):217-225.
[12] 谢田法.两种统计深度函数及相应的中位数[D].北京:中国科学院数学与系统科学研究院,2005.
[13] Marden J.Multivariate ran tests,multivariate analysis,design of experiments,and survey sampling[M].New York:Marcel Dekker,1999:401-432.
[14] Abe S.Support vector machines for pattern classification[M].London:Springer-Verlag London Limited,2005:38-39.
[15] Murphy P M,Aha D W.UCI repository of machine learning databases[EB/OL].[2007].http://www.ics.uci.edu/～mlearn/MLRepository.html.