Bayesian inverse problems with Gaussian priors
Authors:
- B. T. KnapikBKB. T. Knapik
- A. W. van der VaartAVA. W. van der Vaart
- J. H. van ZantenJZJ. H. van Zanten
Open Access
- 1 October 2011
- journal article
- Published by Institute of Mathematical Statistics in The Annals of Statistics
- Vol. 39 (5) , 2626-2657
Abstract
The posterior distribution in a nonparametric inverse problem is shown to contract to the true parameter at a rate that depends on the smoothness of the parameter, and the smoothness and scale of the prior. Correct combinations of these characteristics lead to the minimax rate. The frequentist coverage of credible sets is shown to depend on the combination of prior and true parameter, with smoother priors leading to zero coverage and rougher priors to conservative coverage. In the latter case credible sets are of the correct order of magnitude. The results are numerically illustrated by the problem of recovering a function from observation of a noisy version of its primitive.Keywords
All Related Versions
This publication has 25 references indexed in Scilit:
- On the Bernstein-von Mises phenomenon in the Gaussian white noise modelElectronic Journal of Statistics, 2011
- Nonparametric statistical inverse problemsInverse Problems, 2008
- Bayesian inference with rescaled Gaussian process priorsElectronic Journal of Statistics, 2007
- On adaptive inverse estimation of linear functionals in Hilbert scalesBernoulli, 2003
- Asymptotic Normality of Posterior Distributions in High-Dimensional Linear ModelsBernoulli, 1999
- Statistical Estimation and Optimal RecoveryThe Annals of Statistics, 1994
- An Analysis of Bayesian Inference for Nonparametric RegressionThe Annals of Statistics, 1993
- Renormalization Exponents and Optimal Pointwise Rates of ConvergenceThe Annals of Statistics, 1992
- On Differentiable FunctionalsThe Annals of Statistics, 1991
- Generalized Inverses in Reproducing Kernel Spaces: An Approach to Regularization of Linear Operator EquationsSIAM Journal on Mathematical Analysis, 1974