DocumentCode
1017277
Title
Gradient Projection for Sparse Reconstruction: Application to Compressed Sensing and Other Inverse Problems
Author
Figueiredo, Mário A T ; Nowak, Robert D. ; Wright, Stephen J.
Author_Institution
Inst. de Telecommun., Inst. Superior Tecnico, Lisbon, Portugal
Volume
1
Issue
4
fYear
2007
Firstpage
586
Lastpage
597
Abstract
Many problems in signal processing and statistical inference involve finding sparse solutions to under-determined, or ill-conditioned, linear systems of equations. A standard approach consists in minimizing an objective function which includes a quadratic (squared ) error term combined with a sparseness-inducing regularization term. Basis pursuit, the least absolute shrinkage and selection operator (LASSO), wavelet-based deconvolution, and compressed sensing are a few well-known examples of this approach. This paper proposes gradient projection (GP) algorithms for the bound-constrained quadratic programming (BCQP) formulation of these problems. We test variants of this approach that select the line search parameters in different ways, including techniques based on the Barzilai-Borwein method. Computational experiments show that these GP approaches perform well in a wide range of applications, often being significantly faster (in terms of computation time) than competing methods. Although the performance of GP methods tends to degrade as the regularization term is de-emphasized, we show how they can be embedded in a continuation scheme to recover their efficient practical performance.
Keywords
deconvolution; inverse problems; optimisation; signal reconstruction; statistical analysis; Barzilai-Borwein method; bound-constrained quadratic programming; compressed sensing; convex optimization; gradient projection; inverse problems; least absolute shrinkage selection operator; line search parameters; quadratic error; regularization; signal processing; sparse reconstruction; sparseness-inducing regularization; statistical inference; wavelet-based deconvolution; Compressed sensing; Deconvolution; Degradation; Equations; Inverse problems; Linear systems; Quadratic programming; Signal processing; Signal processing algorithms; Testing; Compressed sensing; convex optimization; deconvolution; gradient projection; quadratic programming; sparse reconstruction; sparseness;
fLanguage
English
Journal_Title
Selected Topics in Signal Processing, IEEE Journal of
Publisher
ieee
ISSN
1932-4553
Type
jour
DOI
10.1109/JSTSP.2007.910281
Filename
4407762
Link To Document