Bayesian Model Selection for Unsupervised Image Deconvolution with Structured Gaussian Priors

B. Harroue; J.-F. Giovannelli; M. Pereyra

doi:10.1109/SSP49050.2021.9513849

Bayesian Model Selection for Unsupervised Image Deconvolution with Structured Gaussian Priors

B. Harroue, J.-F. Giovannelli, M. Pereyra

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

1 Citation (Scopus)

Abstract

This paper considers the comparison of models in the context of inverse problems. Most often, model comparison is addressed in a supervised manner, that can be time-consuming and partly arbitrary. Here we adopt an unsupervised Bayesian approach and quantitatively compare the models based on their posterior probabilities, directly calculated from available data without ground truth available. The probabilities depend on the evidences (marginal likelihoods) of the models and we resort to the Chib approach including a Gibbs sampler to compute them. We focus on the problem of image deconvolution, based on Gaussian models with unknown hyperparameters, in a circulant statement. We compare different impulse responses and covariance structures for image and noise.

Original language	English
Title of host publication	2021 IEEE Statistical Signal Processing Workshop (SSP)
Publisher	IEEE
Pages	241-245
Number of pages	5
ISBN (Electronic)	9781728157672
DOIs	https://doi.org/10.1109/SSP49050.2021.9513849
Publication status	Published - 19 Aug 2021
Event	21st IEEE Statistical Signal Processing Workshop 2021 - Virtual, Rio de Janeiro, Brazil Duration: 11 Jul 2021 → 14 Jul 2021

Conference

Conference	21st IEEE Statistical Signal Processing Workshop 2021
Abbreviated title	SSP 2021
Country/Territory	Brazil
City	Virtual, Rio de Janeiro
Period	11/07/21 → 14/07/21

ASJC Scopus subject areas

Electrical and Electronic Engineering
Applied Mathematics
Signal Processing
Computer Science Applications

Access to Document

10.1109/SSP49050.2021.9513849

Cite this

@inproceedings{27cd56d7f17c4f9285312b48997cf0cd,

title = "Bayesian Model Selection for Unsupervised Image Deconvolution with Structured Gaussian Priors",

abstract = "This paper considers the comparison of models in the context of inverse problems. Most often, model comparison is addressed in a supervised manner, that can be time-consuming and partly arbitrary. Here we adopt an unsupervised Bayesian approach and quantitatively compare the models based on their posterior probabilities, directly calculated from available data without ground truth available. The probabilities depend on the evidences (marginal likelihoods) of the models and we resort to the Chib approach including a Gibbs sampler to compute them. We focus on the problem of image deconvolution, based on Gaussian models with unknown hyperparameters, in a circulant statement. We compare different impulse responses and covariance structures for image and noise. ",

author = "B. Harroue and J.-F. Giovannelli and M. Pereyra",

note = "Publisher Copyright: {\textcopyright} 2021 IEEE.; 21st IEEE Statistical Signal Processing Workshop 2021, SSP 2021 ; Conference date: 11-07-2021 Through 14-07-2021",

year = "2021",

month = aug,

day = "19",

doi = "10.1109/SSP49050.2021.9513849",

language = "English",

pages = "241--245",

booktitle = "2021 IEEE Statistical Signal Processing Workshop (SSP)",

publisher = "IEEE",

address = "United States",

}

TY - GEN

T1 - Bayesian Model Selection for Unsupervised Image Deconvolution with Structured Gaussian Priors

AU - Harroue, B.

AU - Giovannelli, J.-F.

AU - Pereyra, M.

PY - 2021/8/19

Y1 - 2021/8/19

N2 - This paper considers the comparison of models in the context of inverse problems. Most often, model comparison is addressed in a supervised manner, that can be time-consuming and partly arbitrary. Here we adopt an unsupervised Bayesian approach and quantitatively compare the models based on their posterior probabilities, directly calculated from available data without ground truth available. The probabilities depend on the evidences (marginal likelihoods) of the models and we resort to the Chib approach including a Gibbs sampler to compute them. We focus on the problem of image deconvolution, based on Gaussian models with unknown hyperparameters, in a circulant statement. We compare different impulse responses and covariance structures for image and noise.

AB - This paper considers the comparison of models in the context of inverse problems. Most often, model comparison is addressed in a supervised manner, that can be time-consuming and partly arbitrary. Here we adopt an unsupervised Bayesian approach and quantitatively compare the models based on their posterior probabilities, directly calculated from available data without ground truth available. The probabilities depend on the evidences (marginal likelihoods) of the models and we resort to the Chib approach including a Gibbs sampler to compute them. We focus on the problem of image deconvolution, based on Gaussian models with unknown hyperparameters, in a circulant statement. We compare different impulse responses and covariance structures for image and noise.

UR - http://www.scopus.com/inward/record.url?scp=85113520221&partnerID=8YFLogxK

U2 - 10.1109/SSP49050.2021.9513849

DO - 10.1109/SSP49050.2021.9513849

M3 - Conference contribution

AN - SCOPUS:85113520221

SP - 241

EP - 245

BT - 2021 IEEE Statistical Signal Processing Workshop (SSP)

PB - IEEE

T2 - 21st IEEE Statistical Signal Processing Workshop 2021

Y2 - 11 July 2021 through 14 July 2021

ER -

Bayesian Model Selection for Unsupervised Image Deconvolution with Structured Gaussian Priors

Abstract

Conference

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this