Data clustering with cluster size constraints using a modified k-means algorithm

Nuwan Ganganath; Chi Tsun Cheng; Chi Kong Tse

doi:10.1109/CyberC.2014.36

Data clustering with cluster size constraints using a modified k-means algorithm

Nuwan Ganganath, Chi Tsun Cheng, Chi Kong Tse

The Hong Kong Polytechnic University

Research output: Chapter in book / Conference proceeding › Conference article published in proceeding or book › Academic research › peer-review

66 Citations (Scopus)

Abstract

Data clustering is a frequently used technique in finance, computer science, and engineering. In most of the applications, cluster sizes are either constrained to particular values or available as prior knowledge. Unfortunately, traditional clustering methods cannot impose constrains on cluster sizes. In this paper, we propose some vital modifications to the standard k-means algorithm such that it can incorporate size constraints for each cluster separately. The modified k-means algorithm can be used to obtain clusters in preferred sizes. A potential application would be obtaining clusters with equal cluster size. Moreover, the modified algorithm makes use of prior knowledge of the given data set for selectively initializing the cluster centroids which helps escaping from local minima. Simulation results on multidimensional data demonstrate that the k-means algorithm with the proposed modifications can fulfill cluster size constraints and lead to more accurate and robust results.

Original language	English
Title of host publication	Proceedings - 2014 International Conference on Cyber-Enabled Distributed Computing and Knowledge Discovery, CyberC 2014
Publisher	IEEE
Pages	158-161
Number of pages	4
ISBN (Electronic)	9781479962358
DOIs	https://doi.org/10.1109/CyberC.2014.36
Publication status	Published - 1 Jan 2014
Event	6th International Conference on Cyber-Enabled Distributed Computing and Knowledge Discovery, CyberC 2014 - Shanghai, China Duration: 10 Oct 2014 → 12 Oct 2014

Conference

Conference	6th International Conference on Cyber-Enabled Distributed Computing and Knowledge Discovery, CyberC 2014
Country/Territory	China
City	Shanghai
Period	10/10/14 → 12/10/14

Keywords

constrained clustering
data clustering
data mining
k-means
size constraints

ASJC Scopus subject areas

Computational Theory and Mathematics
Computer Science Applications
Software

More information

10.1109/CyberC.2014.36

http://hdl.handle.net/10397/53653

Cite this

@inproceedings{4d96a5b6cd06436db8d1961c05d581e2,

title = "Data clustering with cluster size constraints using a modified k-means algorithm",

abstract = "Data clustering is a frequently used technique in finance, computer science, and engineering. In most of the applications, cluster sizes are either constrained to particular values or available as prior knowledge. Unfortunately, traditional clustering methods cannot impose constrains on cluster sizes. In this paper, we propose some vital modifications to the standard k-means algorithm such that it can incorporate size constraints for each cluster separately. The modified k-means algorithm can be used to obtain clusters in preferred sizes. A potential application would be obtaining clusters with equal cluster size. Moreover, the modified algorithm makes use of prior knowledge of the given data set for selectively initializing the cluster centroids which helps escaping from local minima. Simulation results on multidimensional data demonstrate that the k-means algorithm with the proposed modifications can fulfill cluster size constraints and lead to more accurate and robust results.",

keywords = "constrained clustering, data clustering, data mining, k-means, size constraints",

author = "Nuwan Ganganath and Cheng, {Chi Tsun} and Tse, {Chi Kong}",

year = "2014",

month = jan,

day = "1",

doi = "10.1109/CyberC.2014.36",

language = "English",

pages = "158--161",

booktitle = "Proceedings - 2014 International Conference on Cyber-Enabled Distributed Computing and Knowledge Discovery, CyberC 2014",

publisher = "IEEE",

address = "United States",

note = "6th International Conference on Cyber-Enabled Distributed Computing and Knowledge Discovery, CyberC 2014 ; Conference date: 10-10-2014 Through 12-10-2014",

}

Ganganath, N, Cheng, CT & Tse, CK 2014, Data clustering with cluster size constraints using a modified k-means algorithm. in Proceedings - 2014 International Conference on Cyber-Enabled Distributed Computing and Knowledge Discovery, CyberC 2014., 6984299, IEEE, pp. 158-161, 6th International Conference on Cyber-Enabled Distributed Computing and Knowledge Discovery, CyberC 2014, Shanghai, China, 10/10/14. https://doi.org/10.1109/CyberC.2014.36

Data clustering with cluster size constraints using a modified k-means algorithm. / Ganganath, Nuwan; Cheng, Chi Tsun; Tse, Chi Kong.
Proceedings - 2014 International Conference on Cyber-Enabled Distributed Computing and Knowledge Discovery, CyberC 2014. IEEE, 2014. p. 158-161 6984299.

Research output: Chapter in book / Conference proceeding › Conference article published in proceeding or book › Academic research › peer-review

TY - GEN

T1 - Data clustering with cluster size constraints using a modified k-means algorithm

AU - Ganganath, Nuwan

AU - Cheng, Chi Tsun

AU - Tse, Chi Kong

PY - 2014/1/1

Y1 - 2014/1/1

N2 - Data clustering is a frequently used technique in finance, computer science, and engineering. In most of the applications, cluster sizes are either constrained to particular values or available as prior knowledge. Unfortunately, traditional clustering methods cannot impose constrains on cluster sizes. In this paper, we propose some vital modifications to the standard k-means algorithm such that it can incorporate size constraints for each cluster separately. The modified k-means algorithm can be used to obtain clusters in preferred sizes. A potential application would be obtaining clusters with equal cluster size. Moreover, the modified algorithm makes use of prior knowledge of the given data set for selectively initializing the cluster centroids which helps escaping from local minima. Simulation results on multidimensional data demonstrate that the k-means algorithm with the proposed modifications can fulfill cluster size constraints and lead to more accurate and robust results.

AB - Data clustering is a frequently used technique in finance, computer science, and engineering. In most of the applications, cluster sizes are either constrained to particular values or available as prior knowledge. Unfortunately, traditional clustering methods cannot impose constrains on cluster sizes. In this paper, we propose some vital modifications to the standard k-means algorithm such that it can incorporate size constraints for each cluster separately. The modified k-means algorithm can be used to obtain clusters in preferred sizes. A potential application would be obtaining clusters with equal cluster size. Moreover, the modified algorithm makes use of prior knowledge of the given data set for selectively initializing the cluster centroids which helps escaping from local minima. Simulation results on multidimensional data demonstrate that the k-means algorithm with the proposed modifications can fulfill cluster size constraints and lead to more accurate and robust results.

KW - constrained clustering

KW - data clustering

KW - data mining

KW - k-means

KW - size constraints

UR - http://www.scopus.com/inward/record.url?scp=84921044566&partnerID=8YFLogxK

U2 - 10.1109/CyberC.2014.36

DO - 10.1109/CyberC.2014.36

M3 - Conference article published in proceeding or book

SP - 158

EP - 161

BT - Proceedings - 2014 International Conference on Cyber-Enabled Distributed Computing and Knowledge Discovery, CyberC 2014

PB - IEEE

T2 - 6th International Conference on Cyber-Enabled Distributed Computing and Knowledge Discovery, CyberC 2014

Y2 - 10 October 2014 through 12 October 2014

ER -

Data clustering with cluster size constraints using a modified k-means algorithm

Abstract

Conference

Keywords

ASJC Scopus subject areas

More information

Other files and links

Fingerprint

Cite this