A Bayesian Approach for Multivariate Multifractal Analysis
Speaker: Lorena León (Postdoctoral researcher - CREATIS)
Title: A Bayesian Approach for Multivariate Multifractal Analysis
Abstract: This talk introduces an original Bayesian framework for multivariate multifractal analysis of signals and images. A novel and generic multivariate statistical model is combined with a Whittle-based likelihood approximation and a data augmentation strategy allowing parameter separability. This careful design enables designing efficient estimation procedures for two relevant choices of priors using Gibbs sampling strategies outperforming current methods. The relevance of the proposed framework for modeling real-world data is shown in the important application of drowsiness detection from multichannel physiological signals.