Titre du document / Document title

An improved parameter tuning method for support vector machines

Auteur(s) / Author(s)

YONG QUAN ⁽¹⁾ ; JIE YANG ⁽¹⁾ ;

Affiliation(s) du ou des auteurs / Author(s) Affiliation(s)

⁽¹⁾ Inst. of Image Processing & Pattern Recognition, Shanghai Jiaotong Univ., Shanghai 200030, CHINE

Résumé / Abstract

Support vector machines (SVMs) is a very important tool for data mining. However, the problem of tuning parameters manually limits its application in practical environment. In this paper, under analyzing the limitation of these existing approaches, a new methodology to tuning kernel parameters, based on the computation of the gradient of penalty function with respect to the RBF kernel parameters, is proposed. Simulation results reveal the feasibility of this new approach and demonstrate an improvement of generalization ability.

Revue / Journal Title

Lecture notes in computer science    ISSN  0302-9743 

Source / Source

Congrès
RSFDGrC 2003 : rough sets, fuzzy sets, data mining, and granular computing :   ( Chingqing, 26-29 May 2003 )
Rough sets, fuzzy sets, data mining, and granular computing. International conference N^o9, Chongqing , CHINE (26/05/2003)
2003  , vol. 2639, pp. 607-610[Note(s) : XVII, 741 p., ] [Document : 4 p.] (6 ref.) ISBN 3-540-14040-9 ;  Illustration : Illustration ;

Langue / Language

Anglais

Editeur / Publisher

Springer, Berlin, ALLEMAGNE  (1973) (Revue)
Springer, Berlin, ALLEMAGNE  (2003) (Monographie)

Mots-clés anglais / English Keywords

Vector support machine

;

Penalty function

;

Kernel function

;

Model matching

;

Radial basis function

;

Data mining

;

Information extraction

;

Data analysis

;

Mots-clés français / French Keywords

Machine exemple support

;

Fonction pénalité

;

Fonction noyau

;

Ajustement modèle

;

Fonction base radiale

;

Fouille donnée

;

Extraction information

;

Analyse donnée

;

Mots-clés espagnols / Spanish Keywords

Función penalidad

;

Función núcleo

;

Ajustamiento modelo

;

Función radial base

;

Busca dato

;

Extracción información

;

Análisis datos

;

Localisation / Location

INIST-CNRS, Cote INIST : 16343, 35400010854421.0990