Supervised learning of analysis-sparsity priors with automatic differentiation - Archive ouverte HAL Accéder directement au contenu
Article Dans Une Revue IEEE Signal Processing Letters Année : 2023

Supervised learning of analysis-sparsity priors with automatic differentiation

Résumé

Sparsity priors are commonly used in denoising and image reconstruction. For analysis-type priors, a dictionary defines a representation of signals that is likely to be sparse. In most situations, this dictionary is not known, and is to be recovered from pairs of ground-truth signals and measurements, by minimizing the reconstruction error. This defines a hierarchical optimization problem, which can be cast as a bi-level optimization. Yet, this problem is unsolvable, as reconstructions and their derivative w.r.t. the dictionary have no closed-form expression. However, reconstructions can be iteratively computed using the Forward-Backward splitting (FB) algorithm. In this paper, we approximate reconstructions by the output of the aforementioned FB algorithm. Then, we leverage automatic differentiation to evaluate the gradient of this output w.r.t. the dictionary, which we learn with projected gradient descent. Experiments show that our algorithm successfully learns the 1D Total Variation (TV) dictionary from piecewise constant signals. For the same case study, we propose to constrain our search to dictionaries of 0centered columns, which removes undesired local minima and improves numerical stability.
Fichier principal
Vignette du fichier
learntv.pdf (649.47 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-03518852 , version 1 (10-01-2022)

Licence

Paternité

Identifiants

Citer

Hashem Ghanem, Joseph Salmon, Nicolas Keriven, Samuel Vaiter. Supervised learning of analysis-sparsity priors with automatic differentiation. IEEE Signal Processing Letters, 2023, 30, pp.339-343. ⟨10.1109/LSP.2023.3244511⟩. ⟨hal-03518852⟩
110 Consultations
46 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More