Multifractal Analysis and Statistical study of wavelet leaders : Applications to Mathematical Models and Physiological Data
Overview & Synopsis
Irregular signals are ubiquitous in nature and technology. Multifractal analysis has emerged as a powerful framework to analyze such data. Its goal is to identify and classify the different types of singularities in a function and estimate the size of the sets where these singularities appear. This field originated from the foundational paper by U. Frisch and G. Parisi in 1985, which proposed a relationship between Kolmogorov’s scaling function and the fractal dimensions of the pointwise singularities of the data (multifractal spectrum). This relationship was later reformulated using wavelet decompositions to obtain precise mathematical results concerning the validity of this method.The thesis focuses on several theoretical and applied questions related to multifractal analysis based on wavelet coefficients in an orthonormal basis:- Study of the properties of the weak scaling exponent introduced by Yves Meyer, which allows us to avoid any assumption of global regularity on the data, thus eliminating the need for prior signal regularization by integration.- Validation of this new multifractal formalism on various mathematical models, and its use in the study of physiological data captured from marathon runners, compared to the multifractal formalism based on p-leaders. Joint multifractal analysis (multivariate multifractal analysis) of several physiological signals of different types captured simultaneously.- Theoretical and numerical study of the "leaders" law, which are local suprema of wavelet coefficients and constitute the multiresolution quantities used for performing a multifractal analysis adapted to the pointwise Hölder exponent. The goal of this study is twofold: on one hand, to examine the validity of the log-normality assumption of the coefficients required for the implementation of Bayesian methods for multifractal spectrum estimation; and on the other hand, to lay the groundwork for constructing error bars necessary to obtain classification results based on multifractal parameters. This study leads to the proposal of a more general model based on log-concave distributions, which is numerically validated on both mathematical models and experimental data.- Derivation of precise global and local moduli of continuity for the random wavelet series model, abandoning the assumption of independence of the wavelet coefficients corresponding to different scales. We specifically consider the important case in applications where the distribution of the wavelet coefficients is a mixture of generalized Gaussians.
hal.science)
← / → for pages • + / - for zoom • 0 to fit width
Powered by Mozilla PDF.js v3.11
Document Information
| Document Title | Multifractal Analysis and Statistical study of wavelet leaders : Applications to Mathematical Models and Physiological Data |
|---|---|
| Author / Organization | Wejdene Nasr Ben Hadj Amor |
| Subject & Discipline | Multidisciplinary Academic Open Research |
| Categories | Science & Mathematics |
| Publication Year | 2024 |
| License / Rights | Open Access / Creative Commons |
| Resource Verification | Listed PDF (Verification Pending) |
| Original Host Website |
HAL Open Science
(hal.science)
|
| File Format | application/pdf (Portable Document Format) |
| Language | EN |
| Catalog ID | DOC-245775 |
| Original Source Link | Visit Source Page on hal.science » |