A Bayesian Framework for Multivariate Multifractal Analysis - CNRS - Centre national de la recherche scientifique Accéder directement au contenu
Article Dans Une Revue IEEE Transactions on Signal Processing Année : 2022

A Bayesian Framework for Multivariate Multifractal Analysis

Résumé

Multifractal analysis has become a reference tool for signal and image processing. Grounded in the quantification of local regularity fluctuations, it has proven useful in an increasing range of applications, yet so far involving only univariate data (scalar valued time series or single channel images). Recently the theoretical ground for multivariate multifractal analysis has been devised, showing potential for quantifying transient higher-order dependence beyond linear correlation among collections of data. However, the accurate estimation of the parameters associated with a multivariate multifractal model remains challenging, especially for small sample size data. This work studies an original Bayesian framework for multivariate multifractal estimation, combining a novel and generic multivariate statistical model, a Whittle-based likelihood approximation and a data augmentation strategy allowing parameter separability. This careful design enables efficient estimation procedures to be constructed for two relevant choices of priors using a Gibbs sampling strategy. Monte Carlo simulations, conducted on synthetic multivariate signals and images with various sample sizes and multifractal parameter settings, demonstrate significant performance improvements over the state of the art, at only moderately larger computational cost. Moreover, we show the relevance of the proposed framework for real-world data modeling in the important application of drowsiness detection from multichannel physiological signals.
Fichier principal
Vignette du fichier
BAY-MVMFA-MANUSCRIPT-R2-VF-nocolors.pdf (2.83 Mo) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-03735514 , version 2 (21-07-2022)
hal-03735514 , version 1 (13-11-2022)

Identifiants

Citer

Lorena Leon Arencibia, Herwig Wendt, Jean-Yves Tourneret, Patrice Abry. A Bayesian Framework for Multivariate Multifractal Analysis. IEEE Transactions on Signal Processing, 2022, 70, pp.3663 - 3675. ⟨10.1109/TSP.2022.3187196⟩. ⟨hal-03735514v2⟩
163 Consultations
121 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More