A gradient-like Variational Bayesian approach for inverse scattering problems
Résumé
In this document, we present computations of updating shaping parameters for a new method based on the variational Bayesian approach (VBA) allowing to solve a nonlinear inverse scattering problem. The objective is to detect an unknown object from measurements of the scattered field at different frequencies and for several illuminations. This inverse problem is known to be non-linear and ill-posed. So it needs to be regularized by introducing an a priori information. This is tackled in a Bayesian framework where the particular prior information we account for is that the object is composed of a finite known number of different materials distributed in compact regions. Then we propose the approximate the true joint posterior by a separable law by mean of a gradient-like Variational Bayesian technique. This latter is applied to compute the posterior estimators by allowing a joint update of the shape parameters of the approximating marginals and reconstruct the sought object. The main work is given in [1], while technical details of the variational calculations are presented in the current paper.
Origine : Fichiers produits par l'(les) auteur(s)