titoli, abstracts, parole chiave >>>
Sparseness achievement in Hidden Markov Models

Bicego, Manuele and Cristani, Marco and Murino, Vittorio (2007) Sparseness achievement in Hidden Markov Models. In: Proceedings 14th International Conference on Image Analysis and Processing, 10-14 September 2007, Modena, Italy. Los Alamitos, IEEE Computer Society. p. 67-72. ISBN 978-0-7695-2877-9. Conference or Workshop Item.

Full text disponibile come PDF Richiede visualizzatore di PDF come GSview, Xpdf o Adobe Acrobat Reader

DOI: 10.1109/ICIAP.2007.4362759


In this paper, a novel learning algorithm for Hidden Markov Models (HMMs) has been devised. The key issue is the achievement of a sparse model, i.e., a model in which all irrelevant parameters are set exactly to zero. Alternatively to standard Maximum Likelihood Estimation (Baum Welch training), in the proposed approach the parameters estimation problem is cast into a Bayesian framework, with the introduction of a negative Dirichlet prior, which strongly encourages sparseness of the model. A modified Expectation Maximization algorithm has been devised, able to determine a MAP (Maximum A Posteriori probability) estimate of HMM parameters in this Bayesian formulation. Theoretical considerations and experimental comparative evaluations on a 2D shape classification task contribute to validate the proposed technique.

Item Type:Conference or Workshop Item (Paper)
ID Code:1039
Uncontrolled Keywords:Hidden Markov model, classification
Subjects:Area 09 - Ingegneria industriale e dell'informazione > ING-INF/05 Sistemi di elaborazione delle informazioni
Divisions:001 Università di Sassari > 01 Dipartimenti > Economia, impresa, regolamentazione
Publisher:IEEE Computer Society
Publisher Policy:Depositato in conformità con la politica di copyright dell'Editore
Deposited On:18 Aug 2009 10:04

I documenti depositati in UnissResearch sono protetti dalle leggi che regolano il diritto d'autore

Repository Staff Only: item control page