Dynamic estimation of number of clusters in data sets

被引：29

作者：

Boudraa, AO ^{[1
]}

机构：

[1] Univ Paris 13, Inst Galilee, L2TI, F-93430 Villetaneuse, France

来源：

ELECTRONICS LETTERS | 1999年 / 35卷 / 19期

关键词：

D O I：

10.1049/el:19991151

中图分类号：

TM [电工技术]; TN [电子技术、通信技术];

学科分类号：

0808 ; 0809 ;

摘要：

A new method for estimating during clustering the number of clusters in data sets is proposed. The cluster validity index, B-crit, takes the homogeneity in each cluster into account and is connected to the geometrical properties of the data set. B-crit represents the combination of two validity indices. Comparisons between B-crit and six cluster validity indices, conducted on real data sets, are presented.

引用

页码：1606 / 1608

页数：3

共 12 条

[1]

ANDERSON E, 1939, B AM IRIS SOC, V59, P2

[2] CLUSTER VALIDITY PROFILES [J].

BAILEY, TA ;

DUBES, R .

PATTERN RECOGNITION, 1982, 15 (02) :61-83

[3]

Bezdek J. C, 1975, P 8 INT C NUM TAX SA, P143

[4]

Bezdek JC, 1974, J CYBERNETICS, V3, P58, DOI [10.1080/01969727308546047, DOI 10.1080/01969727308546047]

[5] PRISM - AN ALGORITHM FOR INDUCING MODULAR RULES [J].

CENDROWSKA, J .

INTERNATIONAL JOURNAL OF MAN-MACHINE STUDIES, 1987, 27 (04) :349-370

[6] A SPATIAL-FILTERING APPROACH TO TEXTURE ANALYSIS [J].

COGGINS, JM ;

JAIN, AK .

PATTERN RECOGNITION LETTERS, 1985, 3 (03) :195-203

[7] VALIDITY STUDIES IN CLUSTERING METHODOLOGIES [J].

DUBES, R ;

JAIN, AK .

PATTERN RECOGNITION, 1979, 11 (04) :235-254

[8] HOW MANY CLUSTERS ARE BEST - AN EXPERIMENT [J].

DUBES, RC .

PATTERN RECOGNITION, 1987, 20 (06) :645-663

[9]

Fukuyama Y., 1989, P 5 FUZZ SYST S, V5, P247

[10]

Gunderson R., 1978, P 7 TRIENN WORLD IFA, P1319

← 1 2 →