HAL: hal-00492406, version 1

Detailed view

Export this paper BibTeX - EndNote - TEI - RefWorks -

Available versions:

v1 (2010-06-15)

v2 (2010-09-28)

v3 (2011-01-12)

v4 (2011-04-19)

Simultaneous model-based clustering and visualization in the Fisher discriminative subspace

Charles Bouveyron 1, Camille Brunet 2

(2010-06-15)

Clustering in high-dimensional spaces is nowadays a recurrent problem in many scientific domains but remains a difficult task from both the clustering accuracy and the result understanding points of view. This paper presents a discriminative latent mixture (DLM) model which models the data in a latent orthonormal discriminative subspace with an intrinsic dimension lower than the dimension of the original space. By constraining model parameters within and between groups, a family of 8 parsimonious DLM models is exhibited and this allows to fit onto various situations. An estimation algorithm, called the Fisher-EM algorithm, is also proposed for estimating both the mixture parameters and the discriminative subspace. Experiments on simulated and real datasets show that the proposed approach outperforms existing clustering methods and provides a useful representation of the clustered data. The method is as well applied to the clustering of mass spectrometry data.

1:	Statistique, Analyse et Modélisation Multidisciplinaire (SAmos-Marin Mersenne) (SAMM)
	Université Paris I - Panthéon-Sorbonne
2:	Informatique, Biologie Intégrative et Systèmes Complexes (IBISC)
	CNRS : FRE3190 – Université d'Evry-Val d'Essonne

Subject

:

Mathematics/Statistics Statistics/Statistics Theory

Keyword(s): High-dimensional clustering – Model-based clustering – Discriminative subspace – Fisher criterion – Visualization – Parsimonious models.

hal-00492406, version 1

http://hal-paris1.archives-ouvertes.fr/hal-00492406

oai:hal-paris1.archives-ouvertes.fr:hal-00492406

From: Charles Bouveyron <>

Submitted on: Tuesday, 15 June 2010 17:37:44

Updated on: Tuesday, 15 June 2010 19:37:55