Abstract

As many problems in computer vision, robotics, signal processing and geometric mechanics are expressed as nonsmooth optimization problems on nonlinear spaces (Riemannian manifolds), the huge impact of efficient methods to solve these problems is undeniable. Therefore, it is of eminent interest to develop useful computational and theoretical tools of smooth optimization to nonsmooth problems and of linear spaces to nonlinear spaces. The goal of this talk is a better understanding of theory of nonsmooth analysis on linear spaces and its extension to nonlinear spaces (such as Riemannian manifolds and Hadamard spaces) and considering some of its applications in the real world (for example in image processing)..



Information: