Bernstein-von Mises theorems and uncertainty quantification for linear inverse problems

DOI10.1137/18M1226269MaRDI QIDQ4960994zbMATH OpenOpenAlexFDO

Authors M. Giordano, Hanne Kekkonen

Publication date 24 April 2020

Published in SIAM/ASA Journal on Uncertainty Quantification (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/1811.04058

zbMATH Keywords

asymptotics of nonparametric Bayes proceduresBernstein-von Mises theoremselliptic partial differential equationsGaussian priorsTikhonov regularizers

Mathematics Subject Classification ID

Boundary value problems for second-order elliptic equations (35J25) Heat equation (35K05) Hilbert spaces with reproducing kernels (= (proper) functional Hilbert spaces, including de Branges-Rovnyak and other structured spaces) (46E22) Asymptotic properties of nonparametric inference (62G20) Numerical methods for inverse problems for boundary value problems involving PDEs (65N21)

Abstract: We consider the statistical inverse problem of recovering an unknown function $f$ from a linear measurement corrupted by additive Gaussian white noise. We employ a nonparametric Bayesian approach with standard Gaussian priors, for which the posterior-based reconstruction of $f$ corresponds to a Tikhonov regulariser $$ with a reproducing kernel Hilbert space norm penalty. We prove a semiparametric Bernstein-von Mises theorem for a large collection of linear functionals of $f$, implying that semiparametric posterior estimation and uncertainty quantification are valid and optimal from a frequentist point of view. The result is applied to study three concrete examples that cover both the mildly and severely ill-posed cases: specifically, an elliptic inverse problem, an elliptic boundary value problem and the heat equation. For the elliptic boundary value problem, we also obtain a nonparametric version of the theorem that entails the convergence of the posterior distribution to a prior-independent infinite-dimensional Gaussian probability measure with minimal covariance. As a consequence, it follows that the Tikhonov regulariser $$ is an efficient estimator of $f$, and we derive frequentist guarantees for certain credible balls centred at $$.

Recommendations

• Bernstein-von Mises theorems for statistical inverse problems. I: Schrödinger equation
• Statistical guarantees for Bayesian uncertainty quantification in nonlinear inverse problems with Gaussian process priors
• Regularized posteriors in linear ill-posed inverse problems
• Posterior contraction rates for the Bayesian approach to linear ill-posed inverse problems
• Consistency of Bayesian inference with Gaussian process priors in an elliptic inverse problem

Cites work

• A Bernstein-von Mises theorem for smooth functionals in semiparametric models
• A general approach to posterior contraction in nonparametric inverse problems
• An analysis of Bayesian inference for nonparametric regression
• Analysis of regularized inversion of data corrupted by white Gaussian noise
• Approximation, metric entropy and small ball estimates for Gaussian measures
• Asymptotic equivalence for nonparametric regression with multivariate and random design
• Asymptotic equivalence of nonparametric regression and white noise
• Asymptotic Statistics
• Bayes procedures for adaptive inference in inverse problems for the white noise model
• Bayesian inverse problems with Gaussian priors
• Bayesian inverse problems with non-conjugate priors
• Bayesian inverse problems with partial observations
• Bayesian posterior contraction rates for linear severely ill-posed inverse problems
• Bayesian recovery of the initial condition for the heat equation
• Bernstein-von Mises theorems for statistical inverse problems. I: Schrödinger equation
• Bernstein-von Mises theorems for statistical inverse problems. II: Compound Poisson processes
• Convergence Rates for Penalized Least Squares Estimators in PDE Constrained Regression Problems
• Convergence rates of convex variational regularization
• Discretization-invariant Bayesian inversion and Besov space priors
• Donsker-type theorems for nonparametric maximum likelihood estimators
• Efficient nonparametric Bayesian inference for X-ray transforms
• Frequentist coverage of adaptive nonparametric Bayesian credible sets
• Function spaces and wavelets on domains
• Fundamentals of nonparametric Bayesian inference
• Geometric MCMC for infinite-dimensional inverse problems
• scientific article; zbMATH DE number 1637290 (Why is no real title available?)
• scientific article; zbMATH DE number 3602126 (Why is no real title available?)
• scientific article; zbMATH DE number 1223843 (Why is no real title available?)
• scientific article; zbMATH DE number 1240420 (Why is no real title available?)
• scientific article; zbMATH DE number 1181255 (Why is no real title available?)
• scientific article; zbMATH DE number 954244 (Why is no real title available?)
• scientific article; zbMATH DE number 936298 (Why is no real title available?)
• scientific article; zbMATH DE number 1446717 (Why is no real title available?)
• scientific article; zbMATH DE number 3353865 (Why is no real title available?)
• Inverse problems: a Bayesian perspective
• MAP estimators and their consistency in Bayesian nonparametric inverse problems
• Mathematical foundations of infinite-dimensional statistical models
• Maximum a posteriori probability estimates in infinite-dimensional Bayesian inverse problems
• Non-Gaussian statistical inverse problems. Part I: Posterior distributions
• Nonlinear total variation based noise removal algorithms
• Nonparametric Bernstein-von Mises theorems in Gaussian white noise
• Nonparametric statistical inference for drift vector fields of multi-dimensional diffusions
• On the Bernstein-von Mises phenomenon for nonparametric Bayes procedures
• On the Bernstein-von Mises phenomenon in the Gaussian white noise model
• On the Bernstein-von Mises theorem with infinite-dimensional parameters
• Posterior consistency and convergence rates for Bayesian inversion with hypoelliptic operators
• Posterior consistency for Bayesian inverse problems through stability and regression results
• Posterior contraction rates for the Bayesian approach to linear ill-posed inverse problems
• Rates of contraction of posterior distributions based on Gaussian process priors
• Real Analysis and Probability
• Regularization methods in Banach spaces.
• Regularization theory for ill-posed problems. Selected topics
• Statistical and computational inverse problems.
• Stochastic Equations in Infinite Dimensions
• Uniform Central Limit Theorems

Cited in

(22)

• Statistical guarantees for Bayesian uncertainty quantification in nonlinear inverse problems with Gaussian process priors
• Non-asymptotic error estimates for the Laplace approximation in Bayesian inverse problems
• Nonparametric statistical inference for drift vector fields of multi-dimensional diffusions
• Bernstein-von Mises theorems for statistical inverse problems. II: Compound Poisson processes
• Uncertainty Quantification and Weak Approximation of an Elliptic Inverse Problem
• Continuum limits of posteriors in graph Bayesian inverse problems
• scientific article; zbMATH DE number 5519511 (Why is no real title available?)
• A Bernstein–von-Mises theorem for the Calderón problem with piecewise constant conductivities
• Consistency of Bayesian inference with Gaussian process priors for a parabolic inverse problem
• Solving inverse problems using data-driven models
• Dimension Free Nonasymptotic Bounds on the Accuracy of High-Dimensional Laplace Approximation
• Are Minimizers of the Onsager–Machlup Functional Strong Posterior Modes?
• Convergence Rates for Learning Linear Operators from Noisy Data
• Re-thinking high-dimensional mathematical statistics. Abstracts from the workshop held May 15--21, 2022
• Error control of the numerical posterior with Bayes factors in Bayesian uncertainty quantification
• On some information-theoretic aspects of non-linear statistical inverse problems
• A posterior contraction for Bayesian inverse problems in Banach spaces
• Efficient nonparametric Bayesian inference for X-ray transforms
• Consistency of Bayesian inference for a subdiffusion equation
• Semi-parametric Bernstein-von Mises theorem in linear inverse problems
• Inexact Laplace approximation and the use of posterior mean in Bayesian inference
• Bernstein-von Mises theorems for statistical inverse problems. I: Schrödinger equation

This page was built for publication: Bernstein-von Mises theorems and uncertainty quantification for linear inverse problems

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4960994)