Cluster validation using information stability measures

  • Authors:
  • Damaris Pascual;Filiberto Pla;J. Salvador Sánchez

  • Affiliations:
  • Centro de Reconocimiento de Patrones y Minerıa de Datos, Universidad de Oriente, Av. Patricio Lumumba s/n, Santiago de Cuba 90500, Cuba;Institute of New Imaging Technologies, Departament de Llenguatges i Sistemes Informítics, Universitat Jaume I, 12071 Castelló, Spain;Institute of New Imaging Technologies, Departament de Llenguatges i Sistemes Informítics, Universitat Jaume I, 12071 Castelló, Spain

  • Venue:
  • Pattern Recognition Letters
  • Year:
  • 2010

Quantified Score

Hi-index 0.10

Visualization

Abstract

In this work, a novel technique to address the problem of cluster validation based on cluster stability properties is presented. The stability index here proposed is based on the variation on some information measures over the partitions generated by a given clustering model due to the variability in clustering solutions produced by different sample sets.