Epigraphical Projection and Proximal Tools for Solving Constrained Convex Optimization Problems: Part I - Télécom Paris Accéder directement au contenu
Rapport Année : 2012

Epigraphical Projection and Proximal Tools for Solving Constrained Convex Optimization Problems: Part I

Résumé

We propose a proximal approach to deal with convex optimization problems involving nonlinear constraints. A large family of such constraints, proven to be effective in the solution of inverse problems, can be expressed as the lower level set of a sum of convex functions evaluated over different, but possibly overlapping, blocks of the signal. For this class of constraints, the associated projection operator generally does not have a closed form. We circumvent this difficulty by splitting the lower level set into as many epigraphs as functions involved in the sum. A closed half-space constraint is also enforced, in order to limit the sum of the introduced epigraphical variables to the upper bound of the original lower level set. In this paper, we focus on a family of constraints involving linear transforms of l_1,p balls. Our main theoretical contribution is to provide closed form expressions of the epigraphical projections associated with the Euclidean norm and the sup norm. The proposed approach is validated in the context of image restoration with missing samples, by making use of TV-like constraints. Experiments show that our method leads to significant improvements in term of convergence speed over existing algorithms for solving similar constrained problems.
Fichier principal
Vignette du fichier
hal.pdf (640.25 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)
Loading...

Dates et versions

hal-00744603 , version 1 (23-10-2012)
hal-00744603 , version 2 (15-11-2012)

Identifiants

  • HAL Id : hal-00744603 , version 2

Citer

Giovanni Chierchia, Nelly Pustelnik, Jean-Christophe Pesquet, Béatrice Pesquet-Popescu. Epigraphical Projection and Proximal Tools for Solving Constrained Convex Optimization Problems: Part I. 2012. ⟨hal-00744603v2⟩
364 Consultations
1028 Téléchargements

Partager

Gmail Facebook X LinkedIn More