Unified dual for bi-class SVM approaches

被引：19

作者：

González, L

Angulo, C

Velasco-Morente, F

Català, A

机构：

[1] Univ Seville, Dept Econ Aplicada 1, E-41018 Seville, Spain

[2] Univ Politecn Cataluna, GREC, E-08800 Vilanova I La Geltru, Spain

来源：

PATTERN RECOGNITION | 2005年 / 38卷 / 10期

关键词：

SVM; binary classification; large margin principle; kernels; QP-problem;

D O I：

10.1016/j.patcog.2005.03.019

中图分类号：

TP18 [人工智能理论];

学科分类号：

081104 ; 0812 ; 0835 ; 1405 ;

摘要：

SVM theory was originally developed on the basis of a separable binary classification problem, and other approaches have been later introduced. In this paper, we demonstrated that all these approaches admit the same dual problem formulation. (c) 2005 Pattern Recognition Society. Published by Elsevier Ltd. All rights reserved.

引用

页码：1772 / 1774

页数：3