Efficient stochastic optimisation by unadjusted Langevin Monte Carlo. Application to maximum marginal likelihood and empirical Bayesian estimation

DOI10.1007/S11222-020-09986-YzbMATH Open1475.62026arXiv1906.12281OpenAlexW3109346590MaRDI QIDQ2058738FDOQ2058738

Valentin De Bortoli, Ana F. Vidal, Alain Durmus, Marcelo Pereyra

Publication date: 9 December 2021

Published in: Statistics and Computing (Search for Journal in Brave)

Abstract: Stochastic approximation methods play a central role in maximum likelihood estimation problems involving intractable likelihood functions, such as marginal likelihoods arising in problems with missing or incomplete data, and in parametric empirical Bayesian estimation. Combined with Markov chain Monte Carlo algorithms, these stochastic optimisation methods have been successfully applied to a wide range of problems in science and industry. However, this strategy scales poorly to large problems because of methodological and theoretical difficulties related to using high-dimensional Markov chain Monte Carlo algorithms within a stochastic approximation scheme. This paper proposes to address these difficulties by using unadjusted Langevin algorithms to construct the stochastic approximation. This leads to a highly efficient stochastic optimisation methodology with favourable convergence properties that can be quantified explicitly and easily checked. The proposed methodology is demonstrated with three experiments, including a challenging application to high-dimensional statistical audio analysis and a sparse Bayesian logistic regression with random effects problem.

Full work available at URL: https://arxiv.org/abs/1906.12281

zbMATH Keywords

recursive estimationMarkov chain Monte Carlo methodsstochastic approximationunadjusted Langevin algorithmempirical Bayesian inferencemaximum marginal likelihood estimation

Mathematics Subject Classification ID

Computational methods for problems pertaining to statistics (62-08) Monte Carlo methods (65C05)

Cites Work

• Title not available (Why is that?)
• Title not available (Why is that?)
• Title not available (Why is that?)
• Title not available (Why is that?)
• Title not available (Why is that?)
• Title not available (Why is that?)
• Title not available (Why is that?)
• Title not available (Why is that?)
• Title not available (Why is that?)
• Title not available (Why is that?)
• Exponential convergence of Langevin distributions and their discrete approximations
• Markov chains for exploring posterior distributions. (With discussion)
• Bayesian Inference for Logistic Models Using Pólya–Gamma Latent Variables
• A Stochastic Approximation Method
• Convergence of a stochastic approximation version of the EM algorithm
• Geometric ergodicity of Metropolis algorithms
• Robust Stochastic Approximation Approach to Stochastic Programming
• On the ergodicity properties of some adaptive MCMC algorithms
• Consistency and fluctuations for stochastic gradient Langevin dynamics
• Riemann Manifold Langevin and Hamiltonian Monte Carlo Methods
• Problems and theorems in analysis. I. Series, integral calculus, theory of functions. Transl. from the German by Dorothee Aeppli
• Stability of Markovian processes I: criteria for discrete-time Chains
• Yet Another Look at Harris’ Ergodic Theorem for Markov Chains
• Reflection couplings and contraction rates for diffusions
• On perturbed proximal gradient algorithms
• Convergence of the Monte Carlo expectation maximization for curved exponential families.
• Empirical Bayes: Past, Present and Future
• Handbook of computational statistics. Concepts and methods.
• An introduction to continuous optimization for imaging
• Markov Chains
• Convergence of adaptive and interacting Markov chain Monte Carlo algorithms
• Nonasymptotic convergence analysis for the unadjusted Langevin algorithm
• Theoretical Guarantees for Approximate Sampling from Smooth and Log-Concave Densities
• Couplings and quantitative contraction rates for Langevin dynamics
• On sampling from a log-concave density using kinetic Langevin diffusions
• Quantitative contraction rates for Markov chains on general state spaces
• Exploration of the (non-)asymptotic bias and variance of stochastic gradient Langevin dynamics
• Irreducibility and geometric ergodicity of Hamiltonian Monte Carlo
• Bayesian and frequentist regression methods
• Is there an analog of Nesterov acceleration for gradient-based MCMC?
• Efficient stochastic optimisation by unadjusted Langevin Monte Carlo. Application to maximum marginal likelihood and empirical Bayesian estimation
• Handbook of Convex Optimization Methods in Imaging Science
• 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 I: Methodology and Experiments
• Maximum Likelihood Estimation of Regularization Parameters in High-Dimensional Inverse Problems: An Empirical Bayesian Approach. Part II: Theoretical Analysis

Cited In (6)

• Hybrid unadjusted Langevin methods for high-dimensional latent variable models
• Marginal likelihood estimation in semiblind image deconvolution: a stochastic approximation approach
• Maximum Entropy Methods for Texture Synthesis: Theory and Practice
• Cost free hyper-parameter selection/averaging for Bayesian inverse problems with vanilla and Rao-blackwellized SMC samplers
• The Split Gibbs Sampler Revisited: Improvements to Its Algorithmic Structure and Augmented Target Distribution
• Efficient stochastic optimisation by unadjusted Langevin Monte Carlo. Application to maximum marginal likelihood and empirical Bayesian estimation

This page was built for publication: Efficient stochastic optimisation by unadjusted Langevin Monte Carlo. Application to maximum marginal likelihood and empirical Bayesian estimation

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2058738)