Reconciling "priors" and "priors" without prejudice?

Rémi Gribonval 1 Pierre Machart 1
1 PANAMA - Parcimonie et Nouveaux Algorithmes pour le Signal et la Modélisation Audio
Inria Rennes – Bretagne Atlantique , IRISA-D5 - SIGNAUX ET IMAGES NUMÉRIQUES, ROBOTIQUE
Abstract : There are two major routes to address linear inverse problems. Whereas regularization-based approaches build estimators as solutions of penalized regression optimization problems, Bayesian estimators rely on the posterior distribution of the unknown, given some assumed family of priors. While these may seem radically different approaches, recent results have shown that, in the context of additive white Gaussian denoising, the Bayesian conditional mean estimator is always the solution of a penalized regression problem. The contribution of this paper is twofold. First, we extend the additive white Gaussian denoising results to general linear inverse problems with colored Gaussian noise. Second, we characterize conditions under which the penalty function associated to the conditional mean estimator can satisfy certain popular properties such as convexity, separability, and smoothness. This sheds light on some tradeoff between computational efficiency and estimation accuracy in sparse regularization, and draws some connections between Bayesian estimation and proximal optimization.
Document type :
Conference papers
Liste complète des métadonnées

Cited literature [13 references]  Display  Hide  Download

https://hal.inria.fr/hal-00918008
Contributor : Pierre Machart <>
Submitted on : Thursday, December 12, 2013 - 5:43:00 PM
Last modification on : Friday, November 16, 2018 - 1:40:28 AM
Document(s) archivé(s) le : Friday, March 14, 2014 - 11:40:49 AM

File

NIPS2013_4868.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-00918008, version 1

Citation

Rémi Gribonval, Pierre Machart. Reconciling "priors" and "priors" without prejudice?. Advances in Neural Information Processing Systems, Dec 2013, South Lake Tahoe, United States. ⟨hal-00918008⟩

Share

Metrics

Record views

1011

Files downloads

126