835 articles – 1517 Notices  [english version]
HAL : hal-00414325, version 1

Fiche détaillée  Récupérer au format
Pattern Recognition 43 (2010) 850-858
Parsimonious reduction of Gaussian mixture models with a variational-Bayes approach
Pierrick Bruneau 1, 2, Marc Gelgon 1, 2, Fabien Picarougne 2
(2010)

Aggregating statistical representations of classes is an important task for current trends in scaling up learning and recognition, or for addressing them in distributed in- frastructures. In this perspective, we address the problem of merging probabilistic Gaus- sian mixture models in an efficient way, through the search for a suitable combination of components from mixtures to be merged. We propose a new Bayesian modelling of this combination problem, in association to a variational estimation technique, that handles efficiently the model complexity issue. A main feature of the present scheme is that it merely resorts to the parameters of the original mixture, ensuring low computational cost and possibly communication, should we operate on a distributed system. Experimental results are reported on real data
1 :  ATLAS (INRIA)
INRIA – Université de Nantes
2 :  Laboratoire d'Informatique de Nantes Atlantique (LINA)
CNRS : UMR6241 – Université de Nantes – École Nationale Supérieure des Mines - Nantes
Informatique/Apprentissage

Informatique/Calcul parallèle, distribué et partagé
modèles probabilistes – mélanges de lois – fusion de modèles
Liste des fichiers attachés à ce document : 
PDF
PRpaper-VBEM-HALversion.pdf(289.7 KB)