Optimization Online


Deep Unfolding of a Proximal Interior Point Method for Image Restoration

Carla Bertocchi(carla.bertocchi***at***unimore.it)
Emilie Chouzenoux(emilie.chouzenoux***at***univ-mlv.fr)
Marie-Caroline Corbineau(marie-caroline.corbineau***at***centralesupelec.fr)
Jean-Christophe Pesquet(jean-christophe.pesquet***at***centralesupelec.fr)
Marco Prato(marco.prato***at***unimore.it)

Abstract: Variational methods are widely applied to ill-posed inverse problems for they have the ability to embed prior knowledge about the solution. However, the level of performance of these methods significantly depends on a set of parameters, which can be estimated through computationally expensive and time-consuming methods. In contrast, deep learning offers very generic and efficient architectures, at the expense of explainability, since it is often used as a black-box, without any fine control over its output. Deep unfolding provides a convenient approach to combine variational-based and deep learning approaches. Starting from a variational formulation for image restoration, we develop iRestNet, a neural network architecture obtained by unfolding a proximal interior point algorithm. Hard constraints, encoding desirable properties for the restored image, are incorporated into the network thanks to a logarithmic barrier, while the barrier parameter, the stepsize, and the penalization weight are learned by the network. We derive explicit expressions for the gradient of the proximity operator for various choices of constraints, which allows training iRestNet with gradient descent and backpropagation. In addition, we provide theoretical results regarding the stability of the network for a common inverse problem example. Numerical experiments on image deblurring problems show that the proposed approach compares favorably with both state-of-the-art variational and machine learning methods in terms of image quality.

Keywords: Interior point method, proximal algorithms, deep unfolding, neural network, image restoration, regularization.

Category 1: Convex and Nonsmooth Optimization

Category 2: Robust Optimization

Category 3: Network Optimization

Citation: Scientific report, 12/2018. Universita di Modena e Reggio Emilia, Modena, Italy. CVN, CentraleSupélec, INRIA, Université Paris-Saclay, Gif-Sur-Yvette, France. Université Paris-Est Marne la Vallée, LIGM, UMR CNRS 8049, Champs-sur-Marne, France.

Download: [PDF]

Entry Submitted: 12/11/2018
Entry Accepted: 12/11/2018
Entry Last Modified: 12/11/2018

Modify/Update this entry

  Visitors Authors More about us Links
  Subscribe, Unsubscribe
Digest Archive
Search, Browse the Repository


Coordinator's Board
Classification Scheme
Give us feedback
Optimization Journals, Sites, Societies
Mathematical Optimization Society