A Fast Training Algorithm for Least Squares SVM

Author

Jiang, Shouda ; Lin, Lianlei ; Sun, Chao

Author_Institution

Harbin Inst. of Technol., Harbin

Volume

2

fYear

2007

fDate

26-28 Nov. 2007

Firstpage

586

Lastpage

592

Abstract

A fast training algorithm for Least Squares SVM (LS-SVM) classifiers was proposed, which is based on incremental and decremental learning theory. When a SV (Support Vector) is added or removed, computation based on previous training result replaces large-scale matrix inverse, thus the computation cost is reduced. The innovation is that by reasonable use of incremental and decremental learning the proposed algorithm can adaptively adjust the size of training sets (number of SVs) according to the specific classification problem. Finally several experiments show the validity of proposed algorithm.

Keywords

learning (artificial intelligence); least squares approximations; mathematics computing; pattern classification; support vector machines; decremental learning theory; fast training algorithm; incremental learning theory; least square SVM classifier; support vector machine; Automatic testing; Chaos; Iterative algorithms; Lagrangian functions; Large-scale systems; Least squares methods; Quadratic programming; Sun; Support vector machine classification; Support vector machines;

fLanguage

English

Publisher

ieee

Conference_Titel

Intelligent Information Hiding and Multimedia Signal Processing, 2007. IIHMSP 2007. Third International Conference on

Conference_Location

Kaohsiung

Print_ISBN

978-0-7695-2994-1

Type

conf

DOI

10.1109/IIH-MSP.2007.18

Filename

4457778