Skip to Main content Skip to Navigation
Conference papers

Détermination du nombre de classes dans les méthodes de bipartitionnement

Abstract : Some clustering algorithms require the knowledge of the number of clusters in the popu-lation. The problem becomes more complex with biclustering algorithms where it is necessaryto determine the number of clusters on both dimensions: rows and columns. We propose toextend the application of some indices used for clustering validation to biclustering validation,mainly with Croki2 algorithm. We also propose a new validation index based on the total dif-ferential that we compare with conventional indices. We test the efficiency of these methodson data artificially generated around a number of clusters known on the rows and columns.
Complete list of metadatas

Cited literature [7 references]  Display  Hide  Download

https://hal.archives-ouvertes.fr/hal-01125840
Contributor : Laboratoire Cedric <>
Submitted on : Monday, March 23, 2020 - 9:55:26 AM
Last modification on : Thursday, March 26, 2020 - 1:18:45 AM

File

art_2049.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-01125840, version 1

Citation

Malika Charrad, Yves Lechevallier, Gilbert Saporta, Mohamed Ben Ahmed. Détermination du nombre de classes dans les méthodes de bipartitionnement. 17émes Rencontres de la Société Francophone de Classification, Jun 2010, Saint-Denis de la Réunion, France. pp.119-122. ⟨hal-01125840⟩

Share

Metrics

Record views

158

Files downloads

10