A probabilistic approach to clustering

Research output: Contribution to journalArticlepeer-review

Abstract

The probabilistic approach to clustering, suggested in the paper, is based on examination of an ensemble of clustering problems, in addition to the one given, and estimation of the probability of obtaining clustering of better quality than achieved for the given problem. A Probabilistic Estimate of Clustering is introduced and some of its properties are presented. Some experimental results illustrate the efficiency of the suggested approach.

Original languageEnglish
Pages (from-to)193-198
Number of pages6
JournalPattern Recognition Letters
Volume12
Issue number4
DOIs
StatePublished - Apr 1991

Keywords

  • Cluster
  • Monte-Carlo method
  • cluster validity
  • probabilistic estimate
  • selection of best clustering

Cite this