An improved method to the SVM multi-class classifier based on pairwise coupling

Author

Chen, You ; Zhang, Guo-ji

Author_Institution

Dept. of Appl. Math., South China Univ. of Tech., Guangzhou, China

Volume

5

fYear

2003

fDate

2-5 Nov. 2003

Firstpage

3212

Abstract

In this paper, an improved method is described to solve the SVM multi-class classification problem based on the pairwise coupling. As is described by Zeyu Li and his colleagues, the SVM Multi-class classifier based on pairwise coupling improves the accuracy rate while the computational cost doesn´t increase too much. By the multiple statistical analysis theory, a novel optimal weight matrix is designed to improve the accuracy rate for the less computational cost in this paper. At the end of this paper, the experimental results show the improved method is high-efficient.

Keywords

computational complexity; matrix algebra; pattern classification; pattern clustering; statistical analysis; support vector machines; SVM multiclass classifier; accuracy rate; multiple statistical analysis; optimal weight matrix; pairwise coupling; support vector machine; Computational efficiency; Convergence; Cybernetics; Electronic mail; Machine learning; Statistical analysis; Support vector machine classification; Support vector machines; Time measurement;

fLanguage

English

Publisher

ieee

Conference_Titel

Machine Learning and Cybernetics, 2003 International Conference on

Print_ISBN

0-7803-8131-9

Type

conf

DOI

10.1109/ICMLC.2003.1260133

Filename

1260133