Horseshoe Priors for Edge-Preserving Linear Bayesian Inversion
From MaRDI portal
Abstract: In many large-scale inverse problems, such as computed tomography and image deblurring, characterization of sharp edges in the solution is desired. Within the Bayesian approach to inverse problems, edge-preservation is often achieved using Markov random field priors based on heavy-tailed distributions. Another strategy, popular in statistics, is the application of hierarchical shrinkage priors. An advantage of this formulation lies in expressing the prior as a conditionally Gaussian distribution depending of global and local hyperparameters which are endowed with heavy-tailed hyperpriors. In this work, we revisit the shrinkage horseshoe prior and introduce its formulation for edge-preserving settings. We discuss a sampling framework based on the Gibbs sampler to solve the resulting hierarchical formulation of the Bayesian inverse problem. In particular, one of the conditional distributions is high-dimensional Gaussian, and the rest are derived in closed form by using a scale mixture representation of the heavy-tailed hyperpriors. Applications from imaging science show that our computational procedure is able to compute sharp edge-preserving posterior point estimates with reduced uncertainty.
Cites work
- 10.1162/15324430152748236
- A hybrid Gibbs sampler for edge-preserving tomographic reconstruction with uncertain view angles
- A primer on space-time modeling from a Bayesian perspective
- A stochastic Newton MCMC method for large-scale statistical inverse problems with application to seismic inversion
- AIR tools II: algebraic iterative reconstruction methods, improved implementation
- An introduction to data analysis and uncertainty quantification for inverse problems
- Bayes meets Krylov: statistically inspired preconditioners for CGLS
- Bayesian inference for spatio-temporal spike-and-slab priors
- Bayesian inference with optimal maps
- Bayesian neural network priors for edge-preserving inversion
- Besov priors for Bayesian inverse problems
- Can one use total variation prior for edge-preserving Bayesian inversion?
- Cauchy Markov random field priors for Bayesian inversion
- Computational uncertainty quantification for inverse problems
- Computed tomography. Algorithms, insight, and just enough theory. With contributions by Martin S. Andersen, K. Joost Batenburg, Yiqiu Dond, Eric Todd Quinto and Jan Sijbers
- Deblurring Images
- Discrete inverse problems. Insight and algorithms.
- Discretization-invariant Bayesian inversion and Besov space priors
- From Sparse Solutions of Systems of Equations to Sparse Modeling of Signals and Images
- Gaussian Markov Random Fields
- Geometric ergodicity of Gibbs samplers for the horseshoe and its regularized variants
- Hierachical Bayesian models and sparsity: \(\ell_2\)-magic
- Hierarchical Bayesian level set inversion
- High-dimensional Gaussian sampling: a review and a unifying approach based on a stochastic proximal point algorithm
- scientific article; zbMATH DE number 3513115 (Why is no real title available?)
- scientific article; zbMATH DE number 472922 (Why is no real title available?)
- scientific article; zbMATH DE number 2117879 (Why is no real title available?)
- scientific article; zbMATH DE number 852536 (Why is no real title available?)
- Laplace-distributed increments, the Laplace prior, and edge-preserving regularization
- Markov chain Monte Carlo. Stochastic simulation for Bayesian inference.
- Markov chains for exploring posterior distributions. (With discussion)
- Mean field variational Bayes for elaborate distributions
- On the convergence of the Laplace approximation and noise-level-robustness of Laplace-based Monte Carlo methods for Bayesian inverse problems
- Preconditioned Krylov subspace methods for sampling multivariate Gaussian distributions
- Scalable approximate MCMC algorithms for the horseshoe prior
- Shrinkage priors for Bayesian penalized regression
- Simple conditions for the convergence of the Gibbs sampler and Metropolis-Hastings algorithms
- Sparsity information and regularization in the horseshoe and other shrinkage priors
- Sparsity Promoting Hybrid Solvers for Hierarchical Bayesian Inverse Problems
- Statistical and computational inverse problems.
- Stochastic Relaxation, Gibbs Distributions, and the Bayesian Restoration of Images
- The Bayesian elastic net
- The Bayesian Lasso
- The horseshoe estimator for sparse signals
- The horseshoe estimator: posterior concentration around nearly black vectors
- Well-posed Bayesian inverse problems with infinitely divisible and heavy-tailed prior measures
Cited in
(15)- Can one use total variation prior for edge-preserving Bayesian inversion?
- Sparse Bayesian inference with regularized Gaussian distributions *
- Inducing sparsity via the horseshoe prior in imaging problems
- Certified coordinate selection for high-dimensional Bayesian inversion with Laplace prior
- Adaptive stepsize algorithms for Langevin dynamics
- Bayesian inversion with Student's t priors based on Gaussian scale mixtures
- Uncertainty quantification for linear inverse problems with Besov prior: a randomize-then-optimize method
- Deep Gaussian process priors for Bayesian image reconstruction
- Continuous Gaussian mixture solution for linear Bayesian inversion with application to Laplace priors
- Efficient sparsity-promoting MAP estimation for Bayesian linear inverse problems
- Fused \(L_{1/2}\) prior for large scale linear inverse problem with Gibbs bouncy particle sampler
- Adjoint-Based Bayesian Uncertainty Quantification for PDE-Constrained Inverse Problems with Application to Semiconductor Imaging
- Sparse empirical Bayes with L-smooth hyperpriors
- Efficient sampling for sparse Bayesian learning using hierarchical prior normalization
- Bayesian decomposition using Besov priors
This page was built for publication: Horseshoe Priors for Edge-Preserving Linear Bayesian Inversion
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6156655)
