Training robust support vector regression with smooth non-convex loss function

被引:33
作者
Zhong, Ping [1 ]
机构
[1] China Agr Univ, Coll Sci, Dept Appl Math, Beijing 100083, Peoples R China
基金
中国国家自然科学基金;
关键词
support vector machine; regression; loss function; robustness; d.c; optimization; newton method; FINITE NEWTON METHOD; ALGORITHM; MACHINE; IMPROVEMENTS;
D O I
10.1080/10556788.2011.557725
中图分类号
TP31 [计算机软件];
学科分类号
081202 ; 0835 ;
摘要
The classical support vector machines are constructed based on convex loss functions. Recently, support vector machines with non-convex loss functions have attracted much attention for their superiority to the classical ones in generalization accuracy and robustness. In this paper, we propose a non-convex loss function to construct a robust support vector regression (SVR). The introduced non-convex loss function includes several truncated loss functions as its special cases. The resultant optimization problem is a difference of convex functions program. We employ the concave-convex procedure and develop a Newton-type algorithm to solve it, which can both retain the sparseness of SVR and oppress outliers in the training samples. The experiments on both synthetic and real-world benchmark data sets confirm the robustness and effectiveness of the proposed method.
引用
收藏
页码:1039 / 1058
页数:20
相关论文
共 23 条
[1]   The DC (difference of convex functions) programming and DCA revisited with DC models of real world nonconvex optimization problems [J].
An, LTH ;
Tao, PD .
ANNALS OF OPERATIONS RESEARCH, 2005, 133 (1-4) :23-46
[2]  
[Anonymous], 1997, ACTA MATH VIETNAM
[3]  
[Anonymous], P 23 INT C MACH LEAR
[4]   DC models for spherical separation [J].
Astorino, A. ;
Fuduli, A. ;
Gaudioso, M. .
JOURNAL OF GLOBAL OPTIMIZATION, 2010, 48 (04) :657-669
[5]   Recursive finite Newton algorithm for support vector regression in the primal [J].
Bo, Liefeng ;
Wang, Ling ;
Jiao, Licheng .
NEURAL COMPUTATION, 2007, 19 (04) :1082-1096
[6]  
Chang C.-C., LIBSVM: a Library for Support Vector Machines
[7]   Training a support vector machine in the primal [J].
Chapelle, Olivier .
NEURAL COMPUTATION, 2007, 19 (05) :1155-1178
[8]   Finite Newton method for Lagrangian support vector machine classification [J].
Fung, G ;
Mangasarian, OL .
NEUROCOMPUTING, 2003, 55 (1-2) :39-55
[9]  
Joachims T, 1999, ADVANCES IN KERNEL METHODS, P169
[10]  
Keerthi SS, 2005, J MACH LEARN RES, V6, P341