HAL : hal-00446155, version 1

Fiche détaillée

Récupérer au format BibTeX - EndNote - TEI - RefWorks -

The Ninth IEEE International Conference on Data Mining, Miami, Florida : United States (2009)

A New Clustering Algorithm Based on Regions of Influence with Self-Detection of the Best Number of Clusters

Fabrice Muhlenbach 1, Stéphane Lallich 2

(12/2009)

Clustering methods usually require to know the best number of clusters, or another parameter, e.g. a threshold, which is not ever easy to provide. This paper proposes a new graph-based clustering method called ``GBC'' which detects automatically the best number of clusters, without requiring any other parameter. In this method based on regions of influence, a graph is constructed and the edges of the graph having the higher values are cut according to a hierarchical divisive procedure. An index is calculated from the size average of the cut edges which self-detects the more appropriate number of clusters. The results of GBC for 3 quality indices (Dunn, Silhouette and Davies-Bouldin) are compared with those of K-Means, Ward's hierarchical clustering method and DBSCAN on 8 benchmarks. The experiments show the good performance of GBC in the case of well separated clusters, even if the data are unbalanced, non-convex or with presence of outliers, whatever the shape of the clusters.

1 :	LAboratoire Hubert Curien (LAHC)
	CNRS : UMR5516 – Université Jean Monnet - Saint-Etienne
2 :	Equipe de Recherche en Ingénierie des Connaissances (ERIC)
	Université Lumière - Lyon II : EA3083

Équipe de recherche : Laboratoire Hubert Curien ; laboratoire ERIC

Domaine

:

Informatique/Apprentissage

Mots Clés : clustering – neighborhood graph

hal-00446155, version 1

http://hal.archives-ouvertes.fr/hal-00446155

oai:hal.archives-ouvertes.fr:hal-00446155

Contributeur : Fabrice Muhlenbach <>

Soumis le : Mardi 12 Janvier 2010, 11:03:51

Dernière modification le : Lundi 18 Janvier 2010, 21:07:54