Efficient sampling of non log-concave posterior distributions with mixture of noises

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Palud, Pierre, Thouvenin, Pierre-Antoine, Chainais, Pierre, Bron, Emeric, Petit, Franck Le
Natura: Preprint
Pubblicazione: 2022
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866913786645970944
author Palud, Pierre
Thouvenin, Pierre-Antoine
Chainais, Pierre
Bron, Emeric
Petit, Franck Le
author_facet Palud, Pierre
Thouvenin, Pierre-Antoine
Chainais, Pierre
Bron, Emeric
Petit, Franck Le
contents This paper focuses on a challenging class of inverse problems that is often encountered in applications. The forward model is a complex non-linear black-box, potentially non-injective, whose outputs cover multiple decades in amplitude. Observations are supposed to be simultaneously damaged by additive and multiplicative noises and censorship. As needed in many applications, the aim of this work is to provide uncertainty quantification on top of parameter estimates. The resulting log-likelihood is intractable and potentially non-log-concave. An adapted Bayesian approach is proposed to provide credibility intervals along with point estimates. An MCMC algorithm is proposed to deal with the multimodal posterior distribution, even in a situation where there is no global Lipschitz constant (or it is very large). It combines two kernels, namely an improved version of (Preconditioned Metropolis Adjusted Langevin) PMALA and a Multiple Try Metropolis (MTM) kernel. Whenever smooth, its gradient admits a Lipschitz constant too large to be exploited in the inference process. This sampler addresses all the challenges induced by the complex form of the likelihood. The proposed method is illustrated on classical test multimodal distributions as well as on a challenging and realistic inverse problem in astronomy.
format Preprint
id arxiv_https___arxiv_org_abs_2211_12915
institution arXiv
publishDate 2022
record_format arxiv
spellingShingle Efficient sampling of non log-concave posterior distributions with mixture of noises
Palud, Pierre
Thouvenin, Pierre-Antoine
Chainais, Pierre
Bron, Emeric
Petit, Franck Le
Methodology
This paper focuses on a challenging class of inverse problems that is often encountered in applications. The forward model is a complex non-linear black-box, potentially non-injective, whose outputs cover multiple decades in amplitude. Observations are supposed to be simultaneously damaged by additive and multiplicative noises and censorship. As needed in many applications, the aim of this work is to provide uncertainty quantification on top of parameter estimates. The resulting log-likelihood is intractable and potentially non-log-concave. An adapted Bayesian approach is proposed to provide credibility intervals along with point estimates. An MCMC algorithm is proposed to deal with the multimodal posterior distribution, even in a situation where there is no global Lipschitz constant (or it is very large). It combines two kernels, namely an improved version of (Preconditioned Metropolis Adjusted Langevin) PMALA and a Multiple Try Metropolis (MTM) kernel. Whenever smooth, its gradient admits a Lipschitz constant too large to be exploited in the inference process. This sampler addresses all the challenges induced by the complex form of the likelihood. The proposed method is illustrated on classical test multimodal distributions as well as on a challenging and realistic inverse problem in astronomy.
title Efficient sampling of non log-concave posterior distributions with mixture of noises
topic Methodology
url https://arxiv.org/abs/2211.12915