Skip to Main content Skip to Navigation
Theses

Inférence statistique par des transformées de Fourier pour des modèles de régression semi-paramétriques

Abstract : The shape invariant model consist of the observation of a fixed number of regression functions which differ only by a parametric warping operator. This type of models finds applications in the problems of alignment of continuous signals (images 2D, circadian rhythms,. . .) or discrete (electroencephalogram,. . .). For various warping groups, we propose M-estimators for the parameters characterizing the warping operators associated with the regression functions. These estimators minimize or maximize criteria which are defined with the synchronized average of the Fourier transforms of the data. Moreover, for one of the studied models, we prove the semi-parametric efficiency of the proposed estimator, and we build a test of adequacy of the shape invariant model from one of the criteria.
Document type :
Theses
Complete list of metadatas

https://tel.archives-ouvertes.fr/tel-00185102
Contributor : Myriam Vimond <>
Submitted on : Monday, November 5, 2007 - 1:48:35 PM
Last modification on : Thursday, March 5, 2020 - 5:57:06 PM
Long-term archiving on: : Monday, April 12, 2010 - 1:18:44 AM

Identifiers

  • HAL Id : tel-00185102, version 1

Citation

Myriam Vimond. Inférence statistique par des transformées de Fourier pour des modèles de régression semi-paramétriques. Mathématiques [math]. Université Paul Sabatier - Toulouse III, 2007. Français. ⟨tel-00185102⟩

Share

Metrics

Record views

350

Files downloads

2056