Cluster validation using legacy delineations

Mingqin Liu, Ashok Samal

Research output: Contribution to journalArticle

3 Scopus citations

Abstract

Validation is an important aspect of clustering, but is often overlooked. The existing cluster validity measures are often inadequate for practical applications. Inability to incorporate existing domain knowledge is a primary deficiency. We have proposed a measure which models the domain knowledge in the form of existing delineations, often available from legacy. The effectiveness of the measure in a practical problem is demonstrated.

Original languageEnglish (US)
Pages (from-to)459-467
Number of pages9
JournalImage and Vision Computing
Volume20
Issue number7
DOIs
StatePublished - May 1 2002

Keywords

  • Cluster validation
  • Clustering
  • Fuzzy clustering
  • Legacy datasets

ASJC Scopus subject areas

  • Signal Processing
  • Computer Vision and Pattern Recognition

Cite this