Fast sampling in a linear-Gaussian inverse problem
From MaRDI portal
Publication:3179326
DOI10.1137/15M1029527zbMATH Open1398.94081arXiv1507.01614OpenAlexW2963901148MaRDI QIDQ3179326FDOQ3179326
Authors: Richard A. Norton, Colin Fox 
Publication date: 21 December 2016
Published in: SIAM/ASA Journal on Uncertainty Quantification (Search for Journal in Brave)
Abstract: We solve the inverse problem of deblurring a pixelized image of Jupiter using regularized deconvolution and by sample-based Bayesian inference. By efficiently sampling the marginal posterior distribution for hyperparameters, then the full conditional for the deblurred image, we find that we can evaluate the posterior mean faster than regularized inversion, when selection of the regularizing parameter is considered. To our knowledge, this is the first demonstration of sampling and inference that takes less compute time than regularized inversion in an inverse problems. Comparison to random-walk Metropolis-Hastings and block Gibbs MCMC shows that marginal then conditional sampling also outperforms these more common sampling algorithms, having better scaling with problem size. When problem-specific computations are feasible the asymptotic cost of an independent sample is one linear solve, implying that sample-based Bayesian inference may be performed directly over function spaces, when that limit exists.
Full work available at URL: https://arxiv.org/abs/1507.01614
Recommendations
- Low-rank independence samplers in hierarchical Bayesian inverse problems
- Fast Gibbs sampling for high-dimensional Bayesian inversion
- Fast Markov chain Monte Carlo sampling for sparse Bayesian inference in high-dimensional inverse problems using L1-type priors
- Fast Bayesian inversion for high dimensional inverse problems
- Fast inference for statistical inverse problems
Bayesian inferencehierarchical modeldeconvolutionGibbs samplinglinear inverse problemmarginal algorithmgaussian random fieldmarginal then conditional samplingone-block algorithm
Bayesian inference (62F15) Inverse problems for integral equations (45Q05) Sampling theory in information and communication theory (94A20) Applications of operator theory in probability theory and statistics (47N30)
Cites Work
- Regularization tools: A Matlab package for analysis and solution of discrete ill-posed problems
- Gaussian Markov Random Fields
- Title not available (Why is that?)
- Pattern theory. From representation to inference.
- Spatiotemporal Hierarchical Bayesian Modeling Tropical Ocean Surface Winds
- Title not available (Why is that?)
- The Use of the L-Curve in the Regularization of Discrete Ill-Posed Problems
- A general framework for the parametrization of hierarchical models
- Title not available (Why is that?)
- Radiocarbon Dating with Temporal Order Constraints
- Central limit theorem for additive functionals of reversible Markov processes and applications to simple exclusions
- Title not available (Why is that?)
- An invariant form for the prior probability in estimation problems
- Retrospective Markov chain Monte Carlo methods for Dirichlet process hierarchical models
- Inverse acoustic and electromagnetic scattering theory.
- Numerical Methods in Scientific Computing, Volume I
- Computational Methods for Inverse Problems
- Analysis of the Gibbs Sampler for Hierarchical Inverse Problems
- Title not available (Why is that?)
- Overall objective priors
- Traces and determinants of linear operators
- Harold Jeffreys's \textit{Theory of probability} revisited
- Title not available (Why is that?)
- Title not available (Why is that?)
- Monte Carlo errors with less errors
- MCMC-based image reconstruction with uncertainty quantification
- A primer on space-time modeling from a Bayesian perspective
- Efficient Gaussian Sampling for Solving Large-Scale Inverse Problems Using MCMC
- Estimates of the trace of the inverse of a symmetric matrix using the modified Chebyshev algorithm
- Marginal Markov chain Monte Carlo methods
- A Bayesian linear model for the high-dimensional inverse problem of seismic tomography
Cited In (21)
- Approximation and sampling of multivariate probability distributions in the tensor train decomposition
- Fast sampling of parameterised Gaussian random fields
- Large-Scale Bayesian Spatial-Temporal Regression with Application to Cardiac MR-Perfusion Imaging
- Sampled limited memory methods for massive linear inverse problems
- Polynomial Accelerated Solutions to a Large Gaussian Model for Imaging Biofilms: In Theory and Finite Precision
- Bayesian estimation and uncertainty quantification in models of urea hydrolysis byE. colibiofilms
- Randomized reduced forward models for efficient Metropolis-Hastings MCMC, with application to subsurface fluid flow and capacitance tomography
- Sampling linear inverse problems with noise
- Fast iterative solution of sparsely sampled seismic inverse problems
- Localization for MCMC: sampling high-dimensional posterior distributions with local structure
- Point Spread Function Estimation in X-Ray Imaging with Partially Collapsed Gibbs Sampling
- A blocking scheme for dimension-robust Gibbs sampling in large-scale image deblurring
- Sampling Strategies for Fast Updating of Gaussian Markov Random Fields
- Hybrid iterative ensemble smoother for history matching of hierarchical models
- Low-Rank Independence Samplers in Hierarchical Bayesian Inverse Problems
- Rank Bounds for Approximating Gaussian Densities in the Tensor-Train Format
- A CVAE-within-Gibbs sampler for Bayesian linear inverse problems with hyperparameters
- Optimization-Based Markov Chain Monte Carlo Methods for Nonlinear Hierarchical Statistical Inverse Problems
- Efficient Marginalization-Based MCMC Methods for Hierarchical Bayesian Inverse Problems
- MALA-within-Gibbs Samplers for High-Dimensional Distributions with Sparse Conditional Structure
- Sampling hyperparameters in hierarchical models: Improving on Gibbs for high-dimensional latent fields and large datasets
Uses Software
This page was built for publication: Fast sampling in a linear-Gaussian inverse problem
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3179326)
