EURO-Online login
- New to EURO? Create an account
- I forgot my username and/or my password.
- Help with cookies
(important for IE8 users)
1626. Parameter-free FISTA
Invited abstract in session MB-34: Optimization and learning for data science and imaging (Part II), stream Advances in large scale nonlinear optimization.
Monday, 10:30-12:00Room: 43 (building: 303A)
Authors (first author is the speaker)
| 1. | Luca Calatroni
|
| I3S - CNRS/Université Côte d'Azur | |
| 2. | Jean-François Aujol
|
| IMB, Université de Bordeaux | |
| 3. | Charles Dossal
|
| Université Bordeaux 1 | |
| 4. | Hippolyte Labarrière
|
| Università di Genova | |
| 5. | Aude Rondepierre
|
| Département Génie Mathématiques et Modélisation, INSA Toulouse |
Abstract
We propose an adaptive backtracking and restarting strategy to automate a variant of the Fast Iterative Soft-Thresholding Algorithm (FISTA) for structured optimization problems. Under a generalized (non-restrictive) strong convexity growth condition, we prove the linear convergence of function values generated by the resulting parameter-free algorithm. The algorithm's performance and versatility are demonstrated on exemplar imaging problems.
Keywords
- Non-smooth Optimization
- Computer Science/Applications
- Convex Optimization
Status: accepted
Back to the list of papers