Gradient Scan Gibbs Sampler: An Efficient Algorithm for High-Dimensional Gaussian Distributions. Féron, O., Orieux, F., & Giovannelli, J. IEEE Journal of Selected Topics in Signal Processing, 10(2):343–352, March, 2016. pdf: https://hal-supelec.archives-ouvertes.fr/hal-01252598/file/gsgs-paper.pdfPaper doi abstract bibtex This paper deals with Gibbs samplers that include high dimensional conditional Gaussian distributions. It proposes an efficient algorithm that avoids the high dimensional Gaussian sampling and relies on a random excursion along a small set of directions. The algorithm is proved to converge, i.e., the drawn samples are asymptotically distributed according to the target distribution. Our main motivation is in inverse problems related to general linear observation models and their solution in a hierarchical Bayesian framework implemented through sampling algorithms. It finds direct applications in semi-blind/unsupervised methods as well as in some non-Gaussian methods. The paper provides an illustration focused on the unsupervised estimation for super-resolution methods.
@article{feron_gradient_2016,
title = {Gradient {Scan} {Gibbs} {Sampler}: {An} {Efficient} {Algorithm} for {High}-{Dimensional} {Gaussian} {Distributions}},
volume = {10},
copyright = {All rights reserved},
issn = {1932-4553},
url = {https://hal-supelec.archives-ouvertes.fr/hal-01252598},
doi = {10.1109/JSTSP.2015.2510961},
abstract = {This paper deals with Gibbs samplers that include high dimensional conditional Gaussian distributions. It proposes an efficient algorithm that avoids the high dimensional Gaussian sampling and relies on a random excursion along a small set of directions. The algorithm is proved to converge, i.e., the drawn samples are asymptotically distributed according to the target distribution. Our main motivation is in inverse problems related to general linear observation models and their solution in a hierarchical Bayesian framework implemented through sampling algorithms. It finds direct applications in semi-blind/unsupervised methods as well as in some non-Gaussian methods. The paper provides an illustration focused on the unsupervised estimation for super-resolution methods.},
number = {2},
journal = {IEEE Journal of Selected Topics in Signal Processing},
author = {Féron, Olivier and Orieux, François and Giovannelli, Jean-François},
month = mar,
year = {2016},
note = {pdf: https://hal-supelec.archives-ouvertes.fr/hal-01252598/file/gsgs-paper.pdf},
keywords = {Bayesian, Gaussian, Inverse problems, MCMC, Markov processes, Non-Gaussian, Optimization, Sampling, Signal-Resolution, Super-Resolution, Unsupervised, \_orieux\_publis, general linear observation models},
pages = {343--352},
}
Downloads: 0
{"_id":"EWssYkt6f6muo7LfM","bibbaseid":"fron-orieux-giovannelli-gradientscangibbssampleranefficientalgorithmforhighdimensionalgaussiandistributions-2016","authorIDs":["PWMRGSG6cZehjtbik","mjtQGobtzETGBoZaP"],"author_short":["Féron, O.","Orieux, F.","Giovannelli, J."],"bibdata":{"bibtype":"article","type":"article","title":"Gradient Scan Gibbs Sampler: An Efficient Algorithm for High-Dimensional Gaussian Distributions","volume":"10","copyright":"All rights reserved","issn":"1932-4553","url":"https://hal-supelec.archives-ouvertes.fr/hal-01252598","doi":"10.1109/JSTSP.2015.2510961","abstract":"This paper deals with Gibbs samplers that include high dimensional conditional Gaussian distributions. It proposes an efficient algorithm that avoids the high dimensional Gaussian sampling and relies on a random excursion along a small set of directions. The algorithm is proved to converge, i.e., the drawn samples are asymptotically distributed according to the target distribution. Our main motivation is in inverse problems related to general linear observation models and their solution in a hierarchical Bayesian framework implemented through sampling algorithms. It finds direct applications in semi-blind/unsupervised methods as well as in some non-Gaussian methods. The paper provides an illustration focused on the unsupervised estimation for super-resolution methods.","number":"2","journal":"IEEE Journal of Selected Topics in Signal Processing","author":[{"propositions":[],"lastnames":["Féron"],"firstnames":["Olivier"],"suffixes":[]},{"propositions":[],"lastnames":["Orieux"],"firstnames":["François"],"suffixes":[]},{"propositions":[],"lastnames":["Giovannelli"],"firstnames":["Jean-François"],"suffixes":[]}],"month":"March","year":"2016","note":"pdf: https://hal-supelec.archives-ouvertes.fr/hal-01252598/file/gsgs-paper.pdf","keywords":"Bayesian, Gaussian, Inverse problems, MCMC, Markov processes, Non-Gaussian, Optimization, Sampling, Signal-Resolution, Super-Resolution, Unsupervised, _orieux_publis, general linear observation models","pages":"343–352","bibtex":"@article{feron_gradient_2016,\n\ttitle = {Gradient {Scan} {Gibbs} {Sampler}: {An} {Efficient} {Algorithm} for {High}-{Dimensional} {Gaussian} {Distributions}},\n\tvolume = {10},\n\tcopyright = {All rights reserved},\n\tissn = {1932-4553},\n\turl = {https://hal-supelec.archives-ouvertes.fr/hal-01252598},\n\tdoi = {10.1109/JSTSP.2015.2510961},\n\tabstract = {This paper deals with Gibbs samplers that include high dimensional conditional Gaussian distributions. It proposes an efficient algorithm that avoids the high dimensional Gaussian sampling and relies on a random excursion along a small set of directions. The algorithm is proved to converge, i.e., the drawn samples are asymptotically distributed according to the target distribution. Our main motivation is in inverse problems related to general linear observation models and their solution in a hierarchical Bayesian framework implemented through sampling algorithms. It finds direct applications in semi-blind/unsupervised methods as well as in some non-Gaussian methods. The paper provides an illustration focused on the unsupervised estimation for super-resolution methods.},\n\tnumber = {2},\n\tjournal = {IEEE Journal of Selected Topics in Signal Processing},\n\tauthor = {Féron, Olivier and Orieux, François and Giovannelli, Jean-François},\n\tmonth = mar,\n\tyear = {2016},\n\tnote = {pdf: https://hal-supelec.archives-ouvertes.fr/hal-01252598/file/gsgs-paper.pdf},\n\tkeywords = {Bayesian, Gaussian, Inverse problems, MCMC, Markov processes, Non-Gaussian, Optimization, Sampling, Signal-Resolution, Super-Resolution, Unsupervised, \\_orieux\\_publis, general linear observation models},\n\tpages = {343--352},\n}\n\n","author_short":["Féron, O.","Orieux, F.","Giovannelli, J."],"key":"feron_gradient_2016","id":"feron_gradient_2016","bibbaseid":"fron-orieux-giovannelli-gradientscangibbssampleranefficientalgorithmforhighdimensionalgaussiandistributions-2016","role":"author","urls":{"Paper":"https://hal-supelec.archives-ouvertes.fr/hal-01252598"},"keyword":["Bayesian","Gaussian","Inverse problems","MCMC","Markov processes","Non-Gaussian","Optimization","Sampling","Signal-Resolution","Super-Resolution","Unsupervised","_orieux_publis","general linear observation models"],"metadata":{"authorlinks":{"orieux, f":"https://pro.orieux.fr/"}}},"bibtype":"article","biburl":"https://api.zotero.org/users/5435969/collections/UULHLXA2/items?key=9LGfz4ZAG7Twqg3w0yNxe5EI&format=bibtex&limit=100","creationDate":"2020-06-19T22:47:46.045Z","downloads":0,"keywords":["bayesian","gaussian","inverse problems","mcmc","markov processes","non-gaussian","optimization","sampling","signal-resolution","super-resolution","unsupervised","_orieux_publis","general linear observation models"],"search_terms":["gradient","scan","gibbs","sampler","efficient","algorithm","high","dimensional","gaussian","distributions","féron","orieux","giovannelli"],"title":"Gradient Scan Gibbs Sampler: An Efficient Algorithm for High-Dimensional Gaussian Distributions","year":2016,"dataSources":["rDeciCkFuzBR2ZTtR"]}