K-maximin clustering: A maximin correlation approach to partition-based clustering

Taehoon Lee, Seung Jean Kim, Eui Young Chung, Sungroh Yoon

Research output: Contribution to journalArticle

2 Citations (Scopus)

Abstract

We propose a new clustering algorithm based upon the maximin correlation analysis (MCA), a learning technique that can minimize the maximum misclassification risk. The proposed algorithm resembles conventional partition clustering algorithms such as k-means in that data objects are partitioned into k disjoint partitions. On the other hand, the proposed approach is unique in that an MCA-based approach is used to decide the location of the representative point for each partition. We test the proposed technique with typography data and show our approach outperforms the popular k-means and k-medoids clustering in terms of retrieving the inherent cluster membership.

Original languageEnglish
Pages (from-to)1205-1211
Number of pages7
Journalieice electronics express
Volume6
Issue number17
DOIs
Publication statusPublished - 2009 Sep 10

Fingerprint

Clustering algorithms
partitions
Partitions (building)
learning

All Science Journal Classification (ASJC) codes

  • Electronic, Optical and Magnetic Materials
  • Condensed Matter Physics
  • Electrical and Electronic Engineering

Cite this

Lee, Taehoon ; Kim, Seung Jean ; Chung, Eui Young ; Yoon, Sungroh. / K-maximin clustering : A maximin correlation approach to partition-based clustering. In: ieice electronics express. 2009 ; Vol. 6, No. 17. pp. 1205-1211.
@article{d2ba6e1ca5984baa828005cdc0eed8dc,
title = "K-maximin clustering: A maximin correlation approach to partition-based clustering",
abstract = "We propose a new clustering algorithm based upon the maximin correlation analysis (MCA), a learning technique that can minimize the maximum misclassification risk. The proposed algorithm resembles conventional partition clustering algorithms such as k-means in that data objects are partitioned into k disjoint partitions. On the other hand, the proposed approach is unique in that an MCA-based approach is used to decide the location of the representative point for each partition. We test the proposed technique with typography data and show our approach outperforms the popular k-means and k-medoids clustering in terms of retrieving the inherent cluster membership.",
author = "Taehoon Lee and Kim, {Seung Jean} and Chung, {Eui Young} and Sungroh Yoon",
year = "2009",
month = "9",
day = "10",
doi = "10.1587/elex.6.1205",
language = "English",
volume = "6",
pages = "1205--1211",
journal = "IEICE Electronics Express",
issn = "1349-2543",
publisher = "The Institute of Electronics, Information and Communication Engineers (IEICE)",
number = "17",

}

K-maximin clustering : A maximin correlation approach to partition-based clustering. / Lee, Taehoon; Kim, Seung Jean; Chung, Eui Young; Yoon, Sungroh.

In: ieice electronics express, Vol. 6, No. 17, 10.09.2009, p. 1205-1211.

Research output: Contribution to journalArticle

TY - JOUR

T1 - K-maximin clustering

T2 - A maximin correlation approach to partition-based clustering

AU - Lee, Taehoon

AU - Kim, Seung Jean

AU - Chung, Eui Young

AU - Yoon, Sungroh

PY - 2009/9/10

Y1 - 2009/9/10

N2 - We propose a new clustering algorithm based upon the maximin correlation analysis (MCA), a learning technique that can minimize the maximum misclassification risk. The proposed algorithm resembles conventional partition clustering algorithms such as k-means in that data objects are partitioned into k disjoint partitions. On the other hand, the proposed approach is unique in that an MCA-based approach is used to decide the location of the representative point for each partition. We test the proposed technique with typography data and show our approach outperforms the popular k-means and k-medoids clustering in terms of retrieving the inherent cluster membership.

AB - We propose a new clustering algorithm based upon the maximin correlation analysis (MCA), a learning technique that can minimize the maximum misclassification risk. The proposed algorithm resembles conventional partition clustering algorithms such as k-means in that data objects are partitioned into k disjoint partitions. On the other hand, the proposed approach is unique in that an MCA-based approach is used to decide the location of the representative point for each partition. We test the proposed technique with typography data and show our approach outperforms the popular k-means and k-medoids clustering in terms of retrieving the inherent cluster membership.

UR - http://www.scopus.com/inward/record.url?scp=70349471023&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=70349471023&partnerID=8YFLogxK

U2 - 10.1587/elex.6.1205

DO - 10.1587/elex.6.1205

M3 - Article

AN - SCOPUS:70349471023

VL - 6

SP - 1205

EP - 1211

JO - IEICE Electronics Express

JF - IEICE Electronics Express

SN - 1349-2543

IS - 17

ER -