Sur quelques extensions des chaînes de Markov cachées et couples. Applications à la segmentation non-supervisée de signaux radar.

Abstract : We are interested in the generalization of statistical segmentation procedures based on the hidden Markov chain model, which is classically used in signal processing. We develop 3 main aspects : the introduction of richer dependence structure by the use of pairwise Markov chains, the use of relevant multidimensional statistical models in order to deal with the complex observations obtained with modern radars and statistical estimation of the parameters and computational statistics. In the latter axe, we develop an method based estimating functions and propose a new algorithm that generalizes the classical EM algorithm.
Document type :
Theses
Mathematics [math]. Université Pierre et Marie Curie - Paris VI, 2005. French


https://tel.archives-ouvertes.fr/tel-00011302
Contributor : Nicolas Brunel <>
Submitted on : Thursday, January 5, 2006 - 9:25:27 AM
Last modification on : Thursday, January 5, 2006 - 9:30:13 AM

Identifiers

  • HAL Id : tel-00011302, version 1

Citation

Nicolas Brunel. Sur quelques extensions des chaînes de Markov cachées et couples. Applications à la segmentation non-supervisée de signaux radar.. Mathematics [math]. Université Pierre et Marie Curie - Paris VI, 2005. French. <tel-00011302>

Export

Share

Metrics

Consultation de
la notice

134

Téléchargement du document

501