On convergence properties of the Monte Carlo EM algorithm

DOI10.1214/12-IMSCOLL1003zbMATH Open1329.62287arXiv1206.4768MaRDI QIDQ5499666FDOQ5499666

Authors: Ronald C. Neath 

Publication date: 30 July 2015

Published in: Advances in Modern Statistical Theory and Applications: A Festschrift in honor of Morris L. Eaton (Search for Journal in Brave)

Abstract: The Expectation-Maximization (EM) algorithm (Dempster, Laird and Rubin, 1977) is a popular method for computing maximum likelihood estimates (MLEs) in problems with missing data. Each iteration of the al- gorithm formally consists of an E-step: evaluate the expected complete-data log-likelihood given the observed data, with expectation taken at current pa- rameter estimate; and an M-step: maximize the resulting expression to find the updated estimate. Conditions that guarantee convergence of the EM se- quence to a unique MLE were found by Boyles (1983) and Wu (1983). In complicated models for high-dimensional data, it is common to encounter an intractable integral in the E-step. The Monte Carlo EM algorithm of Wei and Tanner (1990) works around this difficulty by maximizing instead a Monte Carlo approximation to the appropriate conditional expectation. Convergence properties of Monte Carlo EM have been studied, most notably, by Chan and Ledolter (1995) and Fort and Moulines (2003). The goal of this review paper is to provide an accessible but rigorous in- troduction to the convergence properties of EM and Monte Carlo EM. No previous knowledge of the EM algorithm is assumed. We demonstrate the im- plementation of EM and Monte Carlo EM in two simple but realistic examples. We show that if the EM algorithm converges it converges to a stationary point of the likelihood, and that the rate of convergence is linear at best. For Monte Carlo EM we present a readable proof of the main result of Chan and Ledolter (1995), and state without proof the conclusions of Fort and Moulines (2003). An important practical implication of Fort and Moulines's (2003) result relates to the determination of Monte Carlo sample sizes in MCEM; we provide a brief review of the literature (Booth and Hobert, 1999; Caffo, Jank and Jones, 2005) on that problem.

Full work available at URL: https://arxiv.org/abs/1206.4768

Recommendations

• On the simulation size and the convergence of the Monte Carlo EM algorithm via likelihood-based distances
• Conditions for convergence of Monte Carlo EM sequences with an application to product diffusion modeling
• Unequal sampling for Monte Carlo EM algorithms.
• scientific article; zbMATH DE number 5583485
• Ascent-Based Monte Carlo Expectation– Maximization

zbMATH Keywords

EM algorithmmaximum likelihoodMonte Carloconvergencemixed model

Mathematics Subject Classification ID

Point estimation (62F10) Classification and discrimination; cluster analysis (statistical aspects) (62H30) Monte Carlo methods (65C05) Research exposition (monographs, survey articles) pertaining to statistics (62-02)

Cited In (23)

• Convergence properties of the EM algorithm in constrained parameter spaces
• On the Convergence of the Monte Carlo Maximum Likelihood Method for Latent Variable Models
• Inference over radiative transfer models using variational and expectation maximization methods
• Unequal sampling for Monte Carlo EM algorithms.
• An automated (Markov chain) Monte Carlo EM algorithm
• Title not available (Why is that?)
• Theoretical and practical considerations on the convergence properties of the Fisher-EM algorithm
• Ascent-Based Monte Carlo Expectation– Maximization
• A variational approximations-DIC rubric for parameter estimation and mixture model selection within a family setting
• Parameter convergence for EM and MM algorithms
• The ECME algorithm: A simple extension of EM and ECM with faster monotone convergence
• Spatial models for non-Gaussian data with covariate measurement error
• Convergence of the Monte Carlo expectation maximization for curved exponential families.
• Sufficient Conditions for Ergodicity and Convergence of MH, SA, and EM Algorithms
• Conditions for convergence of Monte Carlo EM sequences with an application to product diffusion modeling
• Monte Carlo EM algorithm for multivariate normal distributions under missing data
• Does GEM converge to MLE under Wu's conditions? -- A counter example
• The multiset EM algorithm
• Generalized Linear Mixed Models With Crossed Effects and Unit-specific Survey Weights
• Component-wise Markov chain Monte Carlo: uniform and geometric ergodicity under mixing and composition
• On the convergence rate of random permutation sampler and ECR algorithm in missing data models
• Forward-reverse expectation-maximization algorithm for Markov chains: convergence and numerical analysis
• On the simulation size and the convergence of the Monte Carlo EM algorithm via likelihood-based distances

This page was built for publication: On convergence properties of the Monte Carlo EM algorithm

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5499666)