Skip to Main content Skip to Navigation
Conference papers

Méthode de σ-clipping par point fixe pour l'estimation de la distribution sous H0 dans le cadre de tests multiples

Céline Meillier 1 Raphael Bacher 2, 3 Florent Chatelain 4 Olivier Michel 3
1 MIV - Modèles Images Vision
ICube - Laboratoire des sciences de l'ingénieur, de l'informatique et de l'imagerie
4 GIPSA-SAIGA - GIPSA - Signal et Automatique pour la surveillance, le diagnostic et la biomécanique
GIPSA-DA - Département Automatique, GIPSA-DIS - Département Images et Signal
Abstract : Multiple tests procedures used in different fields such as medical imaging, astronomy, or genetics require the knowledge of the null distribution under H0 which corresponds to the case where the phenomenon of interest is absent, i.e. there is only noise. Some authors as [3, 4, 5] highlight the fact that, even the true H0 samples are not exactly distributed as expected. It is then necessary to estimate the true null distribution directly from the data to correct this misspecification. This ensures more reliable and powerful testing procedures, robust to the possible presence of the signal/phenomenon of interest (H1 alternative hypothesis). In this paper, we propose a new method for mean and variance estimation under H0. We show a comparison of the performances w.r.t. well-known methods of the literature in the case of random gaussian field data whose mean and variance are a priori unknown.
Complete list of metadata

Cited literature [7 references]  Display  Hide  Download
Contributor : Céline Meillier Connect in order to contact the contributor
Submitted on : Tuesday, July 18, 2017 - 1:41:42 PM
Last modification on : Wednesday, November 3, 2021 - 5:24:00 AM
Long-term archiving on: : Saturday, January 27, 2018 - 7:16:03 AM


Files produced by the author(s)


  • HAL Id : hal-01563994, version 1


Céline Meillier, Raphael Bacher, Florent Chatelain, Olivier Michel. Méthode de σ-clipping par point fixe pour l'estimation de la distribution sous H0 dans le cadre de tests multiples. GRETSI 2017 - XXVIème Colloque francophone de traitement du signal et des images, Sep 2017, Juan-Les-Pins, France. ⟨hal-01563994⟩



Record views


Files downloads