Maximum a posteriori probability estimates in infinite-dimensional Bayesian inverse problems

A demanding challenge in Bayesian inversion is to efficiently characterize the posterior distribution. This task is problematic especially in high-dimensional non-Gaussian problems, where the structure of the posterior can be very chaotic and difficult to analyse. Current inverse problem literature often approaches the problem by considering suitable point estimators for the task. Typically the choice is made between the maximum a posteriori (MAP) or the conditional mean (CM) estimate. The benefits of either choice are not well-understood from the perspective of infinite-dimensional theory. Most importantly, there exists no general scheme regarding how to connect the topological description of a MAP estimate to a variational problem. The recent results by Dashti and others (Dashti et al 2013 Inverse Problems 29 095017) resolve this issue for nonlinear inverse problems in Gaussian framework. In this work we improve the current understanding by introducing a novel concept called the weak MAP (wMAP) estimate. We show that any MAP estimate in the sense of Dashti et al (2013 Inverse Problems 29 095017) is a wMAP estimate and, moreover, how the wMAP estimate connects to a variational formulation in general infinite-dimensional non-Gaussian problems. The variational formulation enables to study many properties of the infinite-dimensional MAP estimate that were earlier impossible to study. In a recent work by the authors (Burger and Lucka 2014 Maximum a posteriori estimates in linear inverse problems with logconcave priors are proper bayes estimators preprint) the MAP estimator was studied in the context of the Bayes cost method. Using Bregman distances, proper convex Bayes cost functions were introduced for which the MAP estimator is the Bayes estimator. Here, we generalize these results to the infinite-dimensional setting. Moreover, we discuss the implications of our results for some examples of prior models such as the Besov prior and hierarchical prior.

[1]  R. Ramlau,et al.  A stochastic convergence analysis for Tikhonov regularization with sparsity constraints , 2014 .

[2]  M. Burger,et al.  Maximum a posteriori estimates in linear inverse problems with log-concave priors are proper Bayes estimators , 2014, 1402.5297.

[3]  Dipak K. Dey,et al.  Bayesian model diagnostics using functional Bregman divergence , 2014, J. Multivar. Anal..

[4]  A. Stuart,et al.  MAP estimators and their consistency in Bayesian nonparametric inverse problems , 2013, 1303.4795.

[5]  Sari Lasanen,et al.  Non-Gaussian statistical inverse problems. Part I: Posterior distributions , 2012 .

[6]  V. Kolehmainen,et al.  Sparsity-promoting Bayesian inversion , 2012 .

[7]  Sari Lasanen,et al.  Posterior convergence for approximated unknowns in non-Gaussian statistical inverse problems , 2011, 1112.0906.

[8]  A. Stuart,et al.  Besov priors for Bayesian inverse problems , 2011, 1105.0889.

[9]  V. Bogachev.  Differentiable measures on linear spaces , 2010 .

[10]  Andrew M. Stuart,et al.  Inverse problems: A Bayesian perspective , 2010, Acta Numerica.

[11]  M. Lassas,et al.  Hierarchical models in statistical inverse problems and the Mumford–Shah functional , 2009, 0908.3396.

[12]  Tapio Helin,et al.  On infinite-dimensional hierarchical probability models in statistical inverse problems , 2009, 0907.5322.

[13]  Matti Lassas. Eero Saksman,et al.  Discretization-invariant Bayesian inversion and Besov space priors , 2009, 0901.4220.

[14]  Maya R. Gupta,et al.  Functional Bregman Divergence and Bayesian Estimation of Distributions , 2006, IEEE Transactions on Information Theory.

[15]  V. Bogachev.  Gaussian Measures on a , 2022 .

[16]  M. Hegland.  Approximate Maximum a Posteriori with Gaussian Process Priors , 2007 .

[17]  Faming Liang,et al.  Statistical and Computational Inverse Problems , 2006, Technometrics.

[18]  S. Siltanen,et al.  Can one use total variation prior for edge-preserving Bayesian inversion? , 2004 .

[19]  V. Bogachev,et al.  Absolutely Continuous Flows Generated by Sobolev Class Vector Fields in Finite and Infinite Dimensions , 1999 .

[20]  Vladimir I. Bogachev,et al.  Differentiable measures and the Malliavin calculus , 2010 .

[21]  F. Y. Edgeworth,et al.  The theory of statistics , 1996 .

[22]  M. Lifshits.  Gaussian Random Functions , 1995 .

[23]  Ingrid Daubechies,et al.  Ten Lectures on Wavelets , 1992 .

[24]  V. Bogachev,et al.  Analytic properties of infinite-dimensional distributions , 1990 .

[25]  O G Smolyanov,et al.  Analytic properties of infinite-dimensional distributions , 1990 .

[26]  Erkki Somersalo,et al.  Linear inverse problems for generalised random variables , 1989 .

[27]  E Somersalo,et al.  Linear inverse problems for generalised random variables , 1989 .

[28]  H. Triebel,et al.  Topics in Fourier Analysis and Function Spaces , 1987 .

[29]  A. V. Skorohod,et al.  Integration in Hilbert Space , 1974 .