Regularisation techniques for solving inverse problems
by Samuel Farrens (CEA Saclay)
at Amphi Recherche
In this talk I will introduce the concept of an inverse problem and provide some intuitive examples such as basic linear regression. I will then explain the need for regularisation in the case of ill-posed problems and proceed to focus on two powerful regularisation techniques, namely sparsity and the low-rank approximation. Finally, I will present a specific application of these techniques to the deconvolution of space-based galaxy images with a known spatially varying point spread function. Throughout the talk I will endeavour to highlight the applicability of these mathematical tools to a wide range of problems.