Adaptive sequential Monte Carlo for posterior inference and model selection among complex geological priors

Research output: Contribution to journalArticle

Standard

Adaptive sequential Monte Carlo for posterior inference and model selection among complex geological priors. / Amaya, Macarena; Linde, Niclas; Laloy, Eric.

In: Geophysical Journal International, Vol. 226, 01.08.2021, p. 1220-1238.

Research output: Contribution to journalArticle

Author

Amaya, Macarena ; Linde, Niclas ; Laloy, Eric. / Adaptive sequential Monte Carlo for posterior inference and model selection among complex geological priors. In: Geophysical Journal International. 2021 ; Vol. 226. pp. 1220-1238.

Bibtex - Download

@article{24ef93703e24403a8a4a4f91d4aa50ad,
title = "Adaptive sequential Monte Carlo for posterior inference and model selection among complex geological priors",
abstract = "Bayesian model selection enables comparison and ranking of conceptual subsurface models described by spatial prior models, according to the support provided by available geophysical data. Deep generative neural networks can efficiently encode such complex spatial priors, thereby, allowing for a strong model dimensionality reduction that comes at the price of enhanced non-linearity. In this setting, we explore a recent adaptive sequential Monte Carlo (ASMC) approach that builds on annealed importance sampling (AIS); a method that provides both the posterior probability density function (PDF) and the evidence (a central quantity for Bayesian model selection) through a particle approximation. Both techniques are well suited to parallel computation and rely on importance sampling over a sequence of intermediate distributions, linking the prior and the posterior PDF. Each subsequent distribution is approximated by updating the particle weights and states, compared with the previous approximation, using a small pre-defined number of Markov chain Monte Carlo (MCMC) proposal steps. Compared with AIS, the ASMC method adaptively tunes the tempering between neighboring distributions and performs resampling of particles when the variance of the particle weights becomes too large. We evaluate ASMC using two different conceptual models and associated synthetic cross-hole ground penetrating radar tomography data. For the most challenging test case, we find that the ASMC method is faster and more reliable in locating the posterior PDF than state-of-the-art adaptive MCMC. The evidence estimates are found to be robust with respect to the choice of ASMC algorithmic variables and much less sensitive to the model proposal type than MCMC. The variance of the evidence estimates are best estimated by replication of ASMC runs, while approximations based on single runs provide comparable estimates when using a sufficient number of proposal steps in approximating each intermediate distribution.",
keywords = "Hydrogeophysics, Ground penetrating radar, Inverse theory, Neural networks, Fuzzy logic, Statistical methods, Tomography",
author = "Macarena Amaya and Niclas Linde and Eric Laloy",
note = "Score=10",
year = "2021",
month = "8",
day = "1",
doi = "10.1093/gji/ggab170",
language = "English",
volume = "226",
pages = "1220--1238",
journal = "Geophysical Journal International",
issn = "0956-540X",
publisher = "Oxford University Press",

}

RIS - Download

TY - JOUR

T1 - Adaptive sequential Monte Carlo for posterior inference and model selection among complex geological priors

AU - Amaya, Macarena

AU - Linde, Niclas

AU - Laloy, Eric

N1 - Score=10

PY - 2021/8/1

Y1 - 2021/8/1

N2 - Bayesian model selection enables comparison and ranking of conceptual subsurface models described by spatial prior models, according to the support provided by available geophysical data. Deep generative neural networks can efficiently encode such complex spatial priors, thereby, allowing for a strong model dimensionality reduction that comes at the price of enhanced non-linearity. In this setting, we explore a recent adaptive sequential Monte Carlo (ASMC) approach that builds on annealed importance sampling (AIS); a method that provides both the posterior probability density function (PDF) and the evidence (a central quantity for Bayesian model selection) through a particle approximation. Both techniques are well suited to parallel computation and rely on importance sampling over a sequence of intermediate distributions, linking the prior and the posterior PDF. Each subsequent distribution is approximated by updating the particle weights and states, compared with the previous approximation, using a small pre-defined number of Markov chain Monte Carlo (MCMC) proposal steps. Compared with AIS, the ASMC method adaptively tunes the tempering between neighboring distributions and performs resampling of particles when the variance of the particle weights becomes too large. We evaluate ASMC using two different conceptual models and associated synthetic cross-hole ground penetrating radar tomography data. For the most challenging test case, we find that the ASMC method is faster and more reliable in locating the posterior PDF than state-of-the-art adaptive MCMC. The evidence estimates are found to be robust with respect to the choice of ASMC algorithmic variables and much less sensitive to the model proposal type than MCMC. The variance of the evidence estimates are best estimated by replication of ASMC runs, while approximations based on single runs provide comparable estimates when using a sufficient number of proposal steps in approximating each intermediate distribution.

AB - Bayesian model selection enables comparison and ranking of conceptual subsurface models described by spatial prior models, according to the support provided by available geophysical data. Deep generative neural networks can efficiently encode such complex spatial priors, thereby, allowing for a strong model dimensionality reduction that comes at the price of enhanced non-linearity. In this setting, we explore a recent adaptive sequential Monte Carlo (ASMC) approach that builds on annealed importance sampling (AIS); a method that provides both the posterior probability density function (PDF) and the evidence (a central quantity for Bayesian model selection) through a particle approximation. Both techniques are well suited to parallel computation and rely on importance sampling over a sequence of intermediate distributions, linking the prior and the posterior PDF. Each subsequent distribution is approximated by updating the particle weights and states, compared with the previous approximation, using a small pre-defined number of Markov chain Monte Carlo (MCMC) proposal steps. Compared with AIS, the ASMC method adaptively tunes the tempering between neighboring distributions and performs resampling of particles when the variance of the particle weights becomes too large. We evaluate ASMC using two different conceptual models and associated synthetic cross-hole ground penetrating radar tomography data. For the most challenging test case, we find that the ASMC method is faster and more reliable in locating the posterior PDF than state-of-the-art adaptive MCMC. The evidence estimates are found to be robust with respect to the choice of ASMC algorithmic variables and much less sensitive to the model proposal type than MCMC. The variance of the evidence estimates are best estimated by replication of ASMC runs, while approximations based on single runs provide comparable estimates when using a sufficient number of proposal steps in approximating each intermediate distribution.

KW - Hydrogeophysics

KW - Ground penetrating radar

KW - Inverse theory

KW - Neural networks

KW - Fuzzy logic

KW - Statistical methods

KW - Tomography

UR - https://ecm.sckcen.be/OTCS/llisapi.dll/open/44823930

U2 - 10.1093/gji/ggab170

DO - 10.1093/gji/ggab170

M3 - Article

VL - 226

SP - 1220

EP - 1238

JO - Geophysical Journal International

JF - Geophysical Journal International

SN - 0956-540X

ER -

ID: 7157978