Knowledge-based clustering: A semi-autonomous algorithm using local and global data properties

C. L. Bean, C. Kambhampati

Research output: Contribution to conferencePaper

Abstract

Cluster analysis is a heuristic technique used to reveal inherent groupings in data, but most modern clustering algorithms are highly data and person dependent. This paper presents a clustering technique that minimises the need for user-defined parameters and handles both single and mixed attribute type data sets. The algorithm is based on elements of rough set theory and uses a combination of local and global data properties to obtain meaningful clustering solutions. It is self-consistent in its approach to clustering; thus ensuring the same clustering solution when applied to the same data by different users. The results from a range of real-world and synthetic data sets are used to establish its performance.

Original languageEnglish
Pages95-100
Number of pages6
DOIs
Publication statusPublished - 1 Jan 2004
Externally publishedYes
EventNAFIPS 2004 - Annual Meeting of the North American Fuzzy Information Processing Society: Fuzzy Sets in the Heart of the Canadian Rockies - Banff, Canada
Duration: 27 Jun 200430 Jun 2004

Conference

ConferenceNAFIPS 2004 - Annual Meeting of the North American Fuzzy Information Processing Society
CountryCanada
CityBanff
Period27/06/0430/06/04

Keywords

  • Autonomy
  • Clustering
  • Global
  • Local
  • Rough sets

ASJC Scopus subject areas

  • Computer Science(all)
  • Mathematics(all)

Fingerprint Dive into the research topics of 'Knowledge-based clustering: A semi-autonomous algorithm using local and global data properties'. Together they form a unique fingerprint.

  • Cite this

    Bean, C. L., & Kambhampati, C. (2004). Knowledge-based clustering: A semi-autonomous algorithm using local and global data properties. 95-100. Paper presented at NAFIPS 2004 - Annual Meeting of the North American Fuzzy Information Processing Society, Banff, Canada. https://doi.org/10.1109/nafips.2004.1336256