A hierarchical sparsity-smoothness Bayesian model for ℓ0 + ℓ1 + ℓ2 regularization

Lotfi Chaari; Hadj Batatia; Nicolas Dobigeon; Jean-Yves Tourneret

doi:10.1109/ICASSP.2014.6853929

A hierarchical sparsity-smoothness Bayesian model for ℓ₀ + ℓ₁ + ℓ₂ regularization

Lotfi Chaari, Hadj Batatia, Nicolas Dobigeon, Jean-Yves Tourneret

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

18 Citations (Scopus)

Abstract

Sparse signal/image recovery is a challenging topic that has captured a great interest during the last decades. To address the ill-posedness of the related inverse problem, regularization is often essential by using appropriate priors that promote the sparsity of the target signal/image. In this context, ℓ₀ + ℓ₁ regularization has been widely investigated. In this paper, we introduce a new prior accounting simultaneously for both sparsity and smoothness of restored signals. We use a Bernoulli-generalized Gauss-Laplace distribution to perform ℓ₀ + ℓ₁ + ℓ₂ regularization in a Bayesian framework. Our results show the potential of the proposed approach especially in restoring the non-zero coefficients of the signal/image of interest.

Original language	English
Title of host publication	2014 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)
Publisher	IEEE
Pages	1901-1905
Number of pages	5
ISBN (Electronic)	9781479928934
DOIs	https://doi.org/10.1109/ICASSP.2014.6853929
Publication status	Published - 14 Jul 2014
Event	39th IEEE International Conference on Acoustics, Speech and Signal Processing 2014 - Florence, Italy, Florence, Italy Duration: 4 May 2014 → 9 May 2014 http://www.icassp2014.org/home.html

Publication series

Name	IEEE International Conference on Acoustics, Speech and Signal Processing
ISSN (Print)	1520-6149
ISSN (Electronic)	2379-190X

Conference

Conference	39th IEEE International Conference on Acoustics, Speech and Signal Processing 2014
Abbreviated title	ICASSP 2014
Country/Territory	Italy
City	Florence
Period	4/05/14 → 9/05/14
Other	ICASSP is the world's largest and most comprehensive technical conference focused on signal processing and its applications. The series is sponsored by the IEEE Signal Processing Society and has been held annually since 1976. The conference features world-class speakers, tutorials, exhibits, a Show and Tell event, and over 120 lecture and poster sessions. ICASSP is a cooperative effort of the IEEE Signal Processing Society Technical Committees: Audio and Acoustic Signal Processing Bio Imaging and Signal Processing Design and Implementation of Signal Processing Systems Image, Video, and Multidimenional Signal Processing Information Forensics and Security Industry DSP Technology Standing Committee Machine Learning for Signal Processing Multimedia Signal Processing Sensor Array and Multichannel Signal Processing for Communications and Networking Signal Processing Education Standing Committee Signal Processing Theory and Methods Speech and Langauge Processing
Internet address	http://www.icassp2014.org/home.html

Keywords

hierarchical Bayesian models
MCMC
restoration
smoothness
sparsity

ASJC Scopus subject areas

Software
Signal Processing
Electrical and Electronic Engineering

Access to Document

10.1109/ICASSP.2014.6853929

Cite this

Chaari, L., Batatia, H., Dobigeon, N., & Tourneret, J.-Y. (2014). A hierarchical sparsity-smoothness Bayesian model for ℓ₀ + ℓ₁ + ℓ₂ regularization. In 2014 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) (pp. 1901-1905). (IEEE International Conference on Acoustics, Speech and Signal Processing). IEEE. https://doi.org/10.1109/ICASSP.2014.6853929

@inproceedings{29029ca018654a6ca86805fdc25e920b,

title = "A hierarchical sparsity-smoothness Bayesian model for ℓ0 + ℓ1 + ℓ2 regularization",

abstract = "Sparse signal/image recovery is a challenging topic that has captured a great interest during the last decades. To address the ill-posedness of the related inverse problem, regularization is often essential by using appropriate priors that promote the sparsity of the target signal/image. In this context, ℓ0 + ℓ1 regularization has been widely investigated. In this paper, we introduce a new prior accounting simultaneously for both sparsity and smoothness of restored signals. We use a Bernoulli-generalized Gauss-Laplace distribution to perform ℓ0 + ℓ1 + ℓ2 regularization in a Bayesian framework. Our results show the potential of the proposed approach especially in restoring the non-zero coefficients of the signal/image of interest.",

keywords = "hierarchical Bayesian models, MCMC, restoration, smoothness, sparsity",

author = "Lotfi Chaari and Hadj Batatia and Nicolas Dobigeon and Jean-Yves Tourneret",

year = "2014",

month = jul,

day = "14",

doi = "10.1109/ICASSP.2014.6853929",

language = "English",

series = "IEEE International Conference on Acoustics, Speech and Signal Processing",

publisher = "IEEE",

pages = "1901--1905",

booktitle = "2014 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)",

address = "United States",

note = "39th IEEE International Conference on Acoustics, Speech and Signal Processing 2014 , ICASSP 2014 ; Conference date: 04-05-2014 Through 09-05-2014",

url = "http://www.icassp2014.org/home.html",

}

Chaari, L, Batatia, H, Dobigeon, N & Tourneret, J-Y 2014, A hierarchical sparsity-smoothness Bayesian model for ℓ₀ + ℓ₁ + ℓ₂ regularization. in 2014 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE International Conference on Acoustics, Speech and Signal Processing, IEEE, pp. 1901-1905, 39th IEEE International Conference on Acoustics, Speech and Signal Processing 2014 , Florence, Italy, 4/05/14. https://doi.org/10.1109/ICASSP.2014.6853929

A hierarchical sparsity-smoothness Bayesian model for ℓ₀ + ℓ₁ + ℓ₂ regularization. / Chaari, Lotfi; Batatia, Hadj; Dobigeon, Nicolas et al.
2014 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 2014. p. 1901-1905 (IEEE International Conference on Acoustics, Speech and Signal Processing).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

TY - GEN

T1 - A hierarchical sparsity-smoothness Bayesian model for ℓ0 + ℓ1 + ℓ2 regularization

AU - Chaari, Lotfi

AU - Batatia, Hadj

AU - Dobigeon, Nicolas

AU - Tourneret, Jean-Yves

PY - 2014/7/14

Y1 - 2014/7/14

N2 - Sparse signal/image recovery is a challenging topic that has captured a great interest during the last decades. To address the ill-posedness of the related inverse problem, regularization is often essential by using appropriate priors that promote the sparsity of the target signal/image. In this context, ℓ0 + ℓ1 regularization has been widely investigated. In this paper, we introduce a new prior accounting simultaneously for both sparsity and smoothness of restored signals. We use a Bernoulli-generalized Gauss-Laplace distribution to perform ℓ0 + ℓ1 + ℓ2 regularization in a Bayesian framework. Our results show the potential of the proposed approach especially in restoring the non-zero coefficients of the signal/image of interest.

AB - Sparse signal/image recovery is a challenging topic that has captured a great interest during the last decades. To address the ill-posedness of the related inverse problem, regularization is often essential by using appropriate priors that promote the sparsity of the target signal/image. In this context, ℓ0 + ℓ1 regularization has been widely investigated. In this paper, we introduce a new prior accounting simultaneously for both sparsity and smoothness of restored signals. We use a Bernoulli-generalized Gauss-Laplace distribution to perform ℓ0 + ℓ1 + ℓ2 regularization in a Bayesian framework. Our results show the potential of the proposed approach especially in restoring the non-zero coefficients of the signal/image of interest.

KW - hierarchical Bayesian models

KW - MCMC

KW - restoration

KW - smoothness

KW - sparsity

UR - http://www.scopus.com/inward/record.url?scp=84905233589&partnerID=8YFLogxK

U2 - 10.1109/ICASSP.2014.6853929

DO - 10.1109/ICASSP.2014.6853929

M3 - Conference contribution

AN - SCOPUS:84905233589

T3 - IEEE International Conference on Acoustics, Speech and Signal Processing

SP - 1901

EP - 1905

BT - 2014 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

PB - IEEE

T2 - 39th IEEE International Conference on Acoustics, Speech and Signal Processing 2014

Y2 - 4 May 2014 through 9 May 2014

ER -