Statistical guarantees for Bayesian uncertainty quantification in nonlinear inverse problems with Gaussian process priors

DOI10.1214/21-AOS2082zbMATH Open1486.62068arXiv2007.15892OpenAlexW3046449720MaRDI QIDQ2073706FDOQ2073706

François Monard, Richard Nickl, Gabriel P. Paternain

Publication date: 7 February 2022

Published in: The Annals of Statistics (Search for Journal in Brave)

Abstract: Bayesian inference and uncertainty quantification in a general class of non-linear inverse regression models is considered. Analytic conditions on the regression model $mathscrG(heta):hetainTheta$ and on Gaussian process priors for $heta$ are provided such that semi-parametrically efficient inference is possible for a large class of linear functionals of $heta$. A general semi-parametric Bernstein-von Mises theorem is proved that shows that the (non-Gaussian) posterior distributions are approximated by certain Gaussian measures centred at the posterior mean. As a consequence posterior-based credible sets are valid and optimal from a frequentist point of view. The theory is illustrated with two applications with PDEs that arise in non-linear tomography problems: an elliptic inverse problem for a Schr"odinger equation, and inversion of non-Abelian X-ray transforms. New analytical techniques are deployed to show that the relevant Fisher information operators are invertible between suitable function spaces

Full work available at URL: https://arxiv.org/abs/2007.15892

zbMATH Keywords

[https://portal.mardi4nfdi.de/w/index.php?title=+Special%3ASearch&search=Schr%EF%BF%BD%EF%BF%BDdinger+equation&go=Go Schr��dinger equation]credible setsBernstein-von Mises theoremsX-ray transforms

Mathematics Subject Classification ID

Bayesian inference (62F15) Numerical methods for inverse problems for boundary value problems involving PDEs (65N21)

Cites Work

• Title not available (Why is that?)
• Statistical and computational inverse problems.
• Mathematical Foundations of Infinite-Dimensional Statistical Models
• Asymptotic Statistics
• Real Analysis and Probability
• Whittle-Matérn priors for Bayesian statistical inversion with applications in electrical impedance tomography
• Fundamentals of Nonparametric Bayesian Inference
• Asymptotic methods in statistical decision theory
• Nonparametric statistical inference for drift vector fields of multi-dimensional diffusions
• Bernstein-von Mises theorems for statistical inverse problems. II: Compound Poisson processes
• Nonparametric Bernstein-von Mises theorems in Gaussian white noise
• On the Bernstein-von Mises phenomenon for nonparametric Bayes procedures
• Bernstein-von Mises theorems for statistical inverse problems. I: Schrödinger equation
• A Bernstein-von Mises theorem for smooth functionals in semiparametric models
• Elliptic partial differential equations of second order
• Integral geometry for tensor fields. Transl. from the Russian
• The attenuated ray transform for connections and Higgs fields
• Two dimensional compact simple Riemannian manifolds are boundary distance rigid
• Posterior contraction rates for the Bayesian approach to linear ill-posed inverse problems
• The boundary rigidity problem in the presence of a magnetic field
• Posterior consistency and convergence rates for Bayesian inversion with hypoelliptic operators
• Inverse problems: a Bayesian perspective
• Inverse diffusion theory of photoacoustics
• Bayesian inverse problems with non-conjugate priors
• Fractional Laplacians on domains, a development of Hörmander's theory of \(\mu\)-transmission pseudodifferential operators
• Bayesian inverse problems with Gaussian priors
• The attenuated ray transform on simple surfaces
• On differentiable functionals
• Rates of contraction of posterior distributions based on Gaussian process priors
• Asymptotic equivalence for nonparametric regression with multivariate and random design
• ON DETERMINATION OF A GAUGE FIELD ON ℝ^d FROM ITS NON-ABELIAN RADON TRANSFORM ALONG ORIENTED STRAIGHT LINES
• From Brownian Motion to Schrödinger’s Equation
• Invariant distributions, Beurling transforms and tensor tomography in higher dimensions
• On non-abelian Radon transform
• Inverse boundary spectral problems
• Spectral gaps for a Metropolis-Hastings algorithm in infinite dimensions
• MCMC methods for functions: modifying old algorithms to make them faster
• Statistical inversion and Monte Carlo sampling methods in electrical impedance tomography
• Multi-source quantitative photoacoustic tomography in a diffusive regime
• Complexity analysis of accelerated MCMC methods for Bayesian inversion
• On statistical Calderón problems
• 4. Integral geometry on manifolds with boundary and applications
• Efficient nonparametric Bayesian inference for \(X\)-ray transforms
• Adaptive Bernstein-von Mises theorems in Gaussian white noise
• Consistency of Bayesian inference with Gaussian process priors in an elliptic inverse problem
• Handbook of Uncertainty Quantification
• Convergence Rates for Penalized Least Squares Estimators in PDE Constrained Regression Problems
• Functional Relations, Sharp Mapping Properties, and Regularization of the X-Ray Transform on Disks of Constant Curvature
• Geometric MCMC for infinite-dimensional inverse problems
• Bayesian linear inverse problems in regularity scales
• 5. Non-Abelian Radon transform and its applications
• Bernstein--von Mises Theorems and Uncertainty Quantification for Linear Inverse Problems
• Consistent Inversion of Noisy <scp>Non‐Abelian X‐Ray</scp> Transforms

Cited In (14)

• Nonparametric Bayesian inference for reversible multidimensional diffusions
• Deep learning architectures for nonlinear operator functions and nonlinear inverse problems
• Double b-fibrations and desingularization of the X-ray transform on manifolds with strictly convex boundary
• Adaptive Hessian-Based Nonstationary Gaussian Process Response Surface Method for Probability Density Approximation with Application to Bayesian Solution of Large-Scale Inverse Problems
• The transport Oka-Grauert principle for simple surfaces
• A Bernstein–von-Mises theorem for the Calderón problem with piecewise constant conductivities
• High-dimensional Bernstein-von Mises theorem for the Diaconis-Ylvisaker prior
• On posterior consistency of data assimilation with Gaussian process priors: the 2D-Navier-Stokes equations
• Gaussian process regression in the flat limit
• On some information-theoretic aspects of non-linear statistical inverse problems
• Broken Ray Transform for Twisted Geodesics on Surfaces with a Reflecting Obstacle
• On log-concave approximations of high-dimensional posterior measures and stability properties in non-linear inverse problems
• The C∞ -isomorphism property for a class of singularly-weighted x-ray transforms
• Consistent inference for diffusions from low frequency measurements

This page was built for publication: Statistical guarantees for Bayesian uncertainty quantification in nonlinear inverse problems with Gaussian process priors

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2073706)