Herd Clustering: A synergistic data clustering approach using collective intelligence

Handle URI:
http://hdl.handle.net/10754/563771
Title:
Herd Clustering: A synergistic data clustering approach using collective intelligence
Authors:
Wong, Kachun; Peng, Chengbin ( 0000-0002-7445-2638 ) ; Li, Yue; Chan, Takming
Abstract:
Traditional data mining methods emphasize on analytical abilities to decipher data, assuming that data are static during a mining process. We challenge this assumption, arguing that we can improve the analysis by vitalizing data. In this paper, this principle is used to develop a new clustering algorithm. Inspired by herd behavior, the clustering method is a synergistic approach using collective intelligence called Herd Clustering (HC). The novel part is laid in its first stage where data instances are represented by moving particles. Particles attract each other locally and form clusters by themselves as shown in the case studies reported. To demonstrate its effectiveness, the performance of HC is compared to other state-of-the art clustering methods on more than thirty datasets using four performance metrics. An application for DNA motif discovery is also conducted. The results support the effectiveness of HC and thus the underlying philosophy. © 2014 Elsevier B.V.
KAUST Department:
Computer, Electrical and Mathematical Sciences and Engineering (CEMSE) Division; Computer Science Program
Publisher:
Elsevier BV
Journal:
Applied Soft Computing
Issue Date:
Oct-2014
DOI:
10.1016/j.asoc.2014.05.034
Type:
Article
ISSN:
15684946
Appears in Collections:
Articles; Computer Science Program; Computer, Electrical and Mathematical Sciences and Engineering (CEMSE) Division

Full metadata record

DC FieldValue Language
dc.contributor.authorWong, Kachunen
dc.contributor.authorPeng, Chengbinen
dc.contributor.authorLi, Yueen
dc.contributor.authorChan, Takmingen
dc.date.accessioned2015-08-03T12:09:34Zen
dc.date.available2015-08-03T12:09:34Zen
dc.date.issued2014-10en
dc.identifier.issn15684946en
dc.identifier.doi10.1016/j.asoc.2014.05.034en
dc.identifier.urihttp://hdl.handle.net/10754/563771en
dc.description.abstractTraditional data mining methods emphasize on analytical abilities to decipher data, assuming that data are static during a mining process. We challenge this assumption, arguing that we can improve the analysis by vitalizing data. In this paper, this principle is used to develop a new clustering algorithm. Inspired by herd behavior, the clustering method is a synergistic approach using collective intelligence called Herd Clustering (HC). The novel part is laid in its first stage where data instances are represented by moving particles. Particles attract each other locally and form clusters by themselves as shown in the case studies reported. To demonstrate its effectiveness, the performance of HC is compared to other state-of-the art clustering methods on more than thirty datasets using four performance metrics. An application for DNA motif discovery is also conducted. The results support the effectiveness of HC and thus the underlying philosophy. © 2014 Elsevier B.V.en
dc.publisherElsevier BVen
dc.subjectCollective intelligenceen
dc.subjectHerd behavioren
dc.subjectHeuristicen
dc.subjectNatural computingen
dc.titleHerd Clustering: A synergistic data clustering approach using collective intelligenceen
dc.typeArticleen
dc.contributor.departmentComputer, Electrical and Mathematical Sciences and Engineering (CEMSE) Divisionen
dc.contributor.departmentComputer Science Programen
dc.identifier.journalApplied Soft Computingen
dc.contributor.institutionDepartment of Computer Science, University of Toronto, Toronto, ON, Canadaen
dc.contributor.institutionTerrence Donnelly Centre for Cellular and Biomolecular Research, University of Toronto, Toronto, ON, Canadaen
dc.contributor.institutionDepartment of Integrative Biology and Physiology, University of California Los Angeles, Los Angeles, CA, United Statesen
kaust.authorPeng, Chengbinen
All Items in KAUST are protected by copyright, with all rights reserved, unless otherwise indicated.