Refining pairwise similarity matrix for cluster ensemble problem with cluster relations

Natthakan Iam-On*, Tossapon Boongoen, Simon Garrett

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference Proceeding (Non-Journal item)

66 Citations (SciVal)

Abstract

Cluster ensemble methods have recently emerged as powerful techniques, aggregating several input data clusterings to generate a single output clustering, with improved robustness and stability. This paper presents two new similarity matrices, which are empirically evaluated and compared against the standard co-association matrix on six datasets (both artificial and real data) using four different combination methods and six clustering validity criteria. In all cases, the results suggest the new link-based similarity matrices are able to extract efficiently the information embedded in the input clusterings, and regularly suggest higher clustering quality in comparison to their competitor.

Original languageEnglish
Title of host publicationDiscovery Science - 11th International Conference, DS 2008, Proceedings
EditorsJean-Francois Boulicaut, Michael R. Berthold, Tamás Horváth
PublisherSpringer Nature
Pages222-233
Number of pages12
ISBN (Print)3540884106, 9783540884101
DOIs
Publication statusPublished - 2008
Event11th International Conference on Discovery Science, DS 2008 - Budapest, Hungary
Duration: 13 Oct 200816 Oct 2008

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume5255 LNAI
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference11th International Conference on Discovery Science, DS 2008
Country/TerritoryHungary
CityBudapest
Period13 Oct 200816 Oct 2008

Keywords

  • Cluster ensembles
  • Cluster relation
  • Link analysis
  • Pairwise similarity matrix

Fingerprint

Dive into the research topics of 'Refining pairwise similarity matrix for cluster ensemble problem with cluster relations'. Together they form a unique fingerprint.

Cite this