Cluster validity index based on Jeffrey divergence

Ahmed Ben Said, Rachid Hadjidj, Sebti Foufou

Research output: Contribution to journalArticle

Abstract

Cluster validity indexes are very important tools designed for two purposes: comparing the performance of clustering algorithms and determining the number of clusters that best fits the data. These indexes are in general constructed by combining a measure of compactness and a measure of separation. A classical measure of compactness is the variance. As for separation, the distance between cluster centers is used. However, such a distance does not always reflect the quality of the partition between clusters and sometimes gives misleading results. In this paper, we propose a new cluster validity index for which Jeffrey divergence is used to measure separation between clusters. Experimental results are conducted using different types of data and comparison with widely used cluster validity indexes demonstrates the outperformance of the proposed index.

Original languageEnglish (US)
Pages (from-to)21-31
Number of pages11
JournalPattern Analysis and Applications
Volume20
Issue number1
DOIs
StatePublished - Feb 1 2017

Fingerprint

Clustering algorithms

Keywords

  • Cluster validity index
  • Clustering
  • Jeffrey divergence

ASJC Scopus subject areas

  • Computer Vision and Pattern Recognition
  • Artificial Intelligence

Cite this

Cluster validity index based on Jeffrey divergence. / Said, Ahmed Ben; Hadjidj, Rachid; Foufou, Sebti.

In: Pattern Analysis and Applications, Vol. 20, No. 1, 01.02.2017, p. 21-31.

Research output: Contribution to journalArticle

Said, Ahmed Ben ; Hadjidj, Rachid ; Foufou, Sebti. / Cluster validity index based on Jeffrey divergence. In: Pattern Analysis and Applications. 2017 ; Vol. 20, No. 1. pp. 21-31.
@article{4d0cd67b49e7420598ded34c9c038b2e,
title = "Cluster validity index based on Jeffrey divergence",
abstract = "Cluster validity indexes are very important tools designed for two purposes: comparing the performance of clustering algorithms and determining the number of clusters that best fits the data. These indexes are in general constructed by combining a measure of compactness and a measure of separation. A classical measure of compactness is the variance. As for separation, the distance between cluster centers is used. However, such a distance does not always reflect the quality of the partition between clusters and sometimes gives misleading results. In this paper, we propose a new cluster validity index for which Jeffrey divergence is used to measure separation between clusters. Experimental results are conducted using different types of data and comparison with widely used cluster validity indexes demonstrates the outperformance of the proposed index.",
keywords = "Cluster validity index, Clustering, Jeffrey divergence",
author = "Said, {Ahmed Ben} and Rachid Hadjidj and Sebti Foufou",
year = "2017",
month = "2",
day = "1",
doi = "10.1007/s10044-015-0453-7",
language = "English (US)",
volume = "20",
pages = "21--31",
journal = "Pattern Analysis and Applications",
issn = "1433-7541",
publisher = "Springer London",
number = "1",

}

TY - JOUR

T1 - Cluster validity index based on Jeffrey divergence

AU - Said, Ahmed Ben

AU - Hadjidj, Rachid

AU - Foufou, Sebti

PY - 2017/2/1

Y1 - 2017/2/1

N2 - Cluster validity indexes are very important tools designed for two purposes: comparing the performance of clustering algorithms and determining the number of clusters that best fits the data. These indexes are in general constructed by combining a measure of compactness and a measure of separation. A classical measure of compactness is the variance. As for separation, the distance between cluster centers is used. However, such a distance does not always reflect the quality of the partition between clusters and sometimes gives misleading results. In this paper, we propose a new cluster validity index for which Jeffrey divergence is used to measure separation between clusters. Experimental results are conducted using different types of data and comparison with widely used cluster validity indexes demonstrates the outperformance of the proposed index.

AB - Cluster validity indexes are very important tools designed for two purposes: comparing the performance of clustering algorithms and determining the number of clusters that best fits the data. These indexes are in general constructed by combining a measure of compactness and a measure of separation. A classical measure of compactness is the variance. As for separation, the distance between cluster centers is used. However, such a distance does not always reflect the quality of the partition between clusters and sometimes gives misleading results. In this paper, we propose a new cluster validity index for which Jeffrey divergence is used to measure separation between clusters. Experimental results are conducted using different types of data and comparison with widely used cluster validity indexes demonstrates the outperformance of the proposed index.

KW - Cluster validity index

KW - Clustering

KW - Jeffrey divergence

UR - http://www.scopus.com/inward/record.url?scp=84922388234&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84922388234&partnerID=8YFLogxK

U2 - 10.1007/s10044-015-0453-7

DO - 10.1007/s10044-015-0453-7

M3 - Article

AN - SCOPUS:84922388234

VL - 20

SP - 21

EP - 31

JO - Pattern Analysis and Applications

JF - Pattern Analysis and Applications

SN - 1433-7541

IS - 1

ER -