200. Trust-region methods for relatively smooth weakly convex optimization
Invited abstract in session FD-5: Structured nonsmooth optimization, stream Nonsmooth and nonconvex optimization algorithms.
Friday, 14:10 - 15:50Room: M:N
Authors (first author is the speaker)
| 1. | Mohammad Hamed
|
| Mathematics, University of Antwerpen | |
| 2. | Masoud Ahookhosh
|
| Department of Mathematics, University of Antwerp |
Abstract
Objective functions with relative smoothness (a generalization of the Lipschitz smoothness) and weak convexity (a generalization of convexity) encompass many applications in the domains such as signal and image processing, machine learning and inverse problems. In this talk, we first generate a smooth approximation of the objective function via Bregman forward-backward envelope (BFBE) and design a trust-region method to find a critical point of the original function. This includes, the convergence analysis and the numerical experiments demonstrating the theoretical and practical efficiency of the proposed method.
Keywords
- Convex and non-smooth optimization
Status: accepted
Back to the list of papers