Divergence based feature selection for multimodal class densities

被引：67

作者：

Novovicova, J ^{[1
]}

Pudil, P ^{[1
]}

Kittler, J ^{[1
]}

机构：

[1] UNIV SURREY,DEPT ELECTR & ELECT ENGN,GUILDFORD GU2 5XH,SURREY,ENGLAND

来源：

IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE | 1996年 / 18卷 / 02期

关键词：

feature selection; feature ordering; mixture distribution; maximum likelihood; EM algorithm; Kullback J-divergence;

D O I：

10.1109/34.481557

中图分类号：

TP18 [人工智能理论];

学科分类号：

081104 ; 0812 ; 0835 ; 1405 ;

摘要：

A new feature selection procedure based on the Kullback J-divergence between two class conditional density functions approximated by a finite mixture of parameterized densities of a special type is presented. This procedure is suitable especially for multimodal data. Apart from finding a feature subset of any cardinality without involving any search procedure, it also simultaneously yields a pseudo-Bayes decision rule. Its performance is tested on real data.

引用

页码：218 / 223

页数：6

共 11 条

[1]

[Anonymous], 1982, Pattern recognition: A statistical approach

[2] SOME ASPECTS OF ERROR-BOUNDS IN FEATURE-SELECTION [J].

BOEKEE, DE ;

VANDERLUBBE, JCA .

PATTERN RECOGNITION, 1979, 11 (5-6) :353-360

[3] MAXIMUM LIKELIHOOD FROM INCOMPLETE DATA VIA EM ALGORITHM [J].

DEMPSTER, AP ;

LAIRD, NM ;

RUBIN, DB .

JOURNAL OF THE ROYAL STATISTICAL SOCIETY SERIES B-METHODOLOGICAL, 1977, 39 (01) :1-38

[4]

Duda R. O., 1973, PATTERN CLASSIFICATI, V3

[5]

GRIM J, 1986, KYBERNETIKA, V22, P142

[6]

GRIM J, 1982, KYBERNETIKA, V18, P173

[7]

JAIN AK, 1987, PATTERN RECOGN, P1

[8] FLOATING SEARCH METHODS IN FEATURE-SELECTION [J].

PUDIL, P ;

NOVOVICOVA, J ;

KITTLER, J .

PATTERN RECOGNITION LETTERS, 1994, 15 (11) :1119-1125

[9] SIMULTANEOUS LEARNING OF DECISION RULES AND IMPORTANT ATTRIBUTES FOR CLASSIFICATION PROBLEMS IN IMAGE-ANALYSIS [J].

PUDIL, P ;

NOVOVICOVA, J ;

KITTLER, J .

IMAGE AND VISION COMPUTING, 1994, 12 (03) :193-198

[10]

PUDIL P, 1993, P BMVC 93, V1, P15

← 1 2 →