623. Non-Smooth Optimization on Stiefel Manifold and Beyond
Invited abstract in session WA-1: Plenary 4, stream Plenaries.
Wednesday, 9:00-10:00Room: B100/1001
Authors (first author is the speaker)
| 1. | Anthony So
|
| Department of Systems Engineering and Engineering Management, The Chinese University of Hong Kong |
Abstract
Optimization problems on smooth manifolds often exhibit properties similar to those of unconstrained optimization problems. This has motivated the development of many elegant theoretical tools and efficient numerical methods in manifold optimization over the past decades. Most existing works focus on the setting of optimizing smooth functions on smooth manifolds. In recent years, due to various applications in data science and signal processing, there has been growing interest in developing numerical methods that are provably and practically efficient for optimizing non-smooth functions on smooth manifolds. In this talk, we focus on non-smooth optimization problems on the Stiefel manifold, in which the objective function either has certain composite form or is weakly convex. We present proximal and subgradient-type algorithms for tackling these problems and discuss their convergence properties. We also discuss possible extensions of these developments for non-smooth optimization on other smooth manifolds.
Keywords
- Computational mathematical optimization
Status: accepted
Back to the list of papers