Maximum Likelihood Estimation of Regularization Parameters in High-Dimensional Inverse Problems: An Empirical Bayesian Approach Part I: Methodology and Experiments

DOI10.1137/20M1339829zbMath1457.94014arXiv1911.11709OpenAlexW3100735547MaRDI QIDQ5143322

Ana Fernandez Vidal, Alain Durmus, Marcelo Pereyra, Valentin De Bortoli

Publication date: 11 January 2021

Published in: SIAM Journal on Imaging Sciences (Search for Journal in Brave)

Full work available at URL: https://arxiv.org/abs/1911.11709

zbMATH Keywords

stochastic optimizationempirical Bayesimage processinginverse problemsstatistical inferenceMarkov chain Monte Carlo methodsproximal algorithms

Mathematics Subject Classification ID

Bayesian inference (62F15) Monte Carlo methods (65C05) Computing methodologies for image processing (68U10) Image processing (compression, reconstruction, etc.) in information and communication theory (94A08)

Related Items

A Proximal Markov Chain Monte Carlo Method for Bayesian Inference in Imaging Inverse Problems: When Langevin Meets Moreau, Automatic fidelity and regularization terms selection in variational image restoration, Smoothing unadjusted Langevin algorithms for nonsmooth composite potential functions, The Split Gibbs Sampler Revisited: Improvements to Its Algorithmic Structure and Augmented Target Distribution, On maximum a posteriori estimation with Plug \& Play priors and stochastic gradient descent, Efficient Bayesian Computation for Low-Photon Imaging Problems, On and Beyond Total Variation Regularization in Imaging: The Role of Space Variance, Bayesian Trend Filtering via Proximal Markov Chain Monte Carlo, Maximum Likelihood Estimation of Regularization Parameters in High-Dimensional Inverse Problems: An Empirical Bayesian Approach. Part II: Theoretical Analysis, Efficient stochastic optimisation by unadjusted Langevin Monte Carlo. Application to maximum marginal likelihood and empirical Bayesian estimation, Overcomplete representation in a hierarchical Bayesian framework, Proximal nested sampling for high-dimensional Bayesian model selection

Uses Software

• UNLocBoX

Cites Work

• Unnamed Item
• Unnamed Item
• Unnamed Item
• Unnamed Item
• Bayes procedures for adaptive inference in inverse problems for the white noise model
• The projected GSURE for automatic parameter tuning in iterative shrinkage methods
• A computational framework for empirical Bayes inference
• Concentration of the information in data with log-concave distributions
• Image recovery via total variation minimization and related problems
• Tikhonov regularization and the L-curve for large discrete ill-posed problems
• Statistical and computational inverse problems.
• Risk estimators for choosing regularization parameters in ill-posed problems -- properties and limitations
• A first-order primal-dual algorithm for convex problems with applications to imaging
• Asymptotic behaviour of the empirical Bayes posteriors associated to maximum marginal likelihood estimator
• A discrepancy principle for the Landweber iteration based on risk minimization
• Effective new methods for automated parameter selection in regularized inverse problems
• Stochastic proximal-gradient algorithms for penalized mixed models
• Determining the regularization parameters for super-resolution problems
• An adaptive version for the Metropolis adjusted Langevin algorithm with a truncated drift
• Proximal Markov chain Monte Carlo algorithms
• A Bilevel Optimization Approach for Parameter Learning in Variational Models
• Variational Image Denoising Based on Autocorrelation Whiteness
• Bayes and empirical Bayes: do they merge?
• Proximal Splitting Methods in Signal Processing
• Accelerating Proximal Markov Chain Monte Carlo by Using an Explicit Stabilized Method
• Hypermodels in the Bayesian imaging framework
• Generalized Cross-Validation as a Method for Choosing a Good Ridge Parameter
• The Use of the L-Curve in the Regularization of Discrete Ill-Posed Problems
• 10.1162/15324430152748236
• Asymptotic Statistics
• Bayesian and regularization methods for hyperparameter estimation in image restoration
• Empirical Bayes: Past, Present and Future
• Bilevel approaches for learning of variational imaging models
• A SURE Approach for Digital Signal/Image Deconvolution Problems
• Generalized SURE for Exponential Families: Applications to Regularization
• Handbook of Convex Optimization Methods in Imaging Science
• Hierachical Bayesian models and sparsity: ℓ ₂ -magic
• Efficient Bayesian Computation by Proximal Markov Chain Monte Carlo: When Langevin Meets Moreau
• Maximum Likelihood Estimation of Regularization Parameters in High-Dimensional Inverse Problems: An Empirical Bayesian Approach. Part II: Theoretical Analysis
• Stein Unbiased GrAdient estimator of the Risk (SUGAR) for Multiple Parameter Selection
• Sparse reconstructions from few noisy data: analysis of hierarchical Bayesian models with generalized gamma hyperpriors
• Solving inverse problems using data-driven models
• The Bayesian Choice
• Learning optimal spatially-dependent regularization parameters in total variation image denoising
• On perturbed proximal gradient algorithms
• Fast Image Recovery Using Variable Splitting and Constrained Optimization
• Variational Bayesian Super Resolution
• Estimating the Granularity Coefficient of a Potts-Markov Random Field Within a Markov Chain Monte Carlo Algorithm
• Parameter Estimation for Blind and Non-Blind Deblurring Using Residual Whiteness Measures
• An introduction to continuous optimization for imaging
• Total Generalized Variation
• Convex analysis and monotone operator theory in Hilbert spaces
• Maximum Likelihood Estimation of Misspecified Models
• Bayesian computation: a summary of the current state, and samples backwards and forwards