Analysis Operator Learning for Overcomplete Cosparse Representations - Inria - Institut national de recherche en sciences et technologies du numérique Access content directly
Conference Papers Year : 2011

Analysis Operator Learning for Overcomplete Cosparse Representations

Abstract

We consider the problem of learning a low-dimensional signal model from a collection of training samples. The mainstream approach would be to learn an overcomplete dictionary to provide good approximations of the training samples using sparse synthesis coefficients. This famous sparse model has a less well known counterpart, in analysis form, called the cosparse analysis model. In this new model, signals are characterized by their parsimony in a transformed domain using an overcomplete analysis operator. We propose to learn an analysis operator from a training corpus using a constrained optimization program based on L1 optimization. We derive a practical learning algorithm, based on projected subgradients, and demonstrate its ability to robustly recover a ground truth analysis operator, provided the training set is of sufficient size. A local optimality condition is derived, providing preliminary theoretical support for the well-posedness of the learning problem under appropriate conditions.
Fichier principal
Vignette du fichier
YNGD_CAOL_EUSIPCO11-2.pdf (168.86 Ko) Télécharger le fichier
Origin : Files produced by the author(s)
Loading...

Dates and versions

inria-00583133 , version 1 (02-07-2011)

Identifiers

  • HAL Id : inria-00583133 , version 1

Cite

Mehrdad Yaghoobi, Sangnam Nam, Rémi Gribonval, Mike E. Davies. Analysis Operator Learning for Overcomplete Cosparse Representations. European Signal Processing Conference (EUSIPCO'11), Aug 2011, Barcelona, Spain. ⟨inria-00583133⟩
828 View
554 Download

Share

Gmail Facebook X LinkedIn More