Online EM Algorithm for Latent Data Models

Abstract

In this contribution, we propose a generic online (also sometimes called adaptive or recursive) version of the Expectation-Maximisation (EM) algorithm applicable to latent variable models of independent observations. Compared to the algorithm of Titterington (1984), this approach is more directly connected to the usual EM algorithm and does not rely on integration with respect to the complete data distribution. The resulting algorithm is usually simpler and is shown to achieve convergence to the stationary points of the Kullback-Leibler divergence between the marginal distribution of the observation and the model distribution at the optimal rate, i.e., that of the maximum likelihood estimator. In addition, the proposed approach is also suitable for conditional (or regression) models, as illustrated in the case of the mixture of linear regressions model.

Version that includes the corrigendum published in volume 73, part 5 (2011), of the Journal of the Royal Statistical Society, Series B + the correction of a typo in Eqs. (32-33)

References in corpus (1)

• Convergence rate and averaging of nonlinear two-time-scale stochastic approximation algorithms

Cited by in corpus (71)

• Machine learning in acoustics: theory and applications
• On Particle Methods for Parameter Estimation in State-Space Models
• Multi-Center Federated Learning: Clients Clustering for Better Personalization
• Multi-Center Federated Learning: Clients Clustering for Better Personalization
• Stochastic Successive Convex Approximation for Non-Convex Constrained Stochastic Optimization
• Sparse Stochastic Inference for Latent Dirichlet allocation
• Stochastic Majorization-Minimization Algorithms for Large-Scale Optimization
• Joint segmentation of multivariate time series with hidden process regression for human activity recognition
• Distributed Maximum Likelihood for Simultaneous Self-localization and Tracking in Sensor Networks
• A Probabilistic Theory of Deep Learning
• Learning Hierarchical Shape Segmentation and Labeling from Online Repositories
• HiGrad: Uncertainty Quantification for Online Learning and Stochastic Approximation
• A Mixture-Based Approach to Regional Adaptation for MCMC
• Non-asymptotic Analysis of Biased Stochastic Approximation Scheme
• Fast Online EM for Big Topic Modeling
• BLOB : A Probabilistic Model for Recommendation that Combines Organic and Bandit Signals
• Expectation-maximization for logistic regression
• Inertial Stochastic PALM (iSPALM) and Applications in Machine Learning
• Scalable Bayesian Non-Negative Tensor Factorization for Massive Count Data
• Gaussian Mixture Models for Affordance Learning using Bayesian Networks
• Optimal noise estimation from syndrome statistics of quantum codes
• An approximate expectation-maximization for two-dimensional multi-target detection
• End-to-end Learning of a Fisher Vector Encoding for Part Features in Fine-grained Recognition
• Online Expectation-Maximization Based Frequency and Phase Consensus in Distributed Phased Arrays
• Unbiased Learning to Rank with Biased Continuous Feedback
• Deterministic Approximate EM Algorithm; Application to the Riemann Approximation EM and the Tempered EM
• Neural Density Estimation and Likelihood-free Inference
• Unbiased Learning to Rank: Online or Offline?
• Sinkhorn EM: An Expectation-Maximization algorithm based on entropic optimal transport
• The Bayesian Learning Rule
• Online and Distributed learning of Gaussian mixture models by Bayesian Moment Matching
• Towards Big Topic Modeling
• Convergence Analysis of Riemannian Stochastic Approximation Schemes
• Latent Tree Models for Hierarchical Topic Detection
• Task Recommendation in Crowdsourcing Based on Learning Preferences and Reliabilities
• Online EM Algorithm for Hidden Markov Models
• Divide-and-Conquer Bayesian Inference in Hidden Markov Models
• Bootstrapping Robotic Ecological Perception from a Limited Set of Hypotheses Through Interactive Perception
• Divergence-Based Motivation for Online EM and Combining Hidden Variable Models
• A Stochastic Path-Integrated Differential EstimatoR Expectation Maximization Algorithm
• Closed-form Marginal Likelihood in Gamma-Poisson Matrix Factorization
• Online Inference for Mixture Model of Streaming Graph Signals with Non-White Excitation
• Online Robust and Adaptive Learning from Data Streams
• Maximum likelihood estimation of a finite mixture of logistic regression models in a continuous data stream
• On Riemannian Stochastic Approximation Schemes with Fixed Step-Size
• Scalable Extreme Deconvolution
• On-line Spot Volatility-Estimation and Decomposition with Nonlinear Market Microstructure Noise Models
• Scalable Bayesian reduced-order models for high-dimensional multiscale dynamical systems
• Domain adaptation for sequence labeling using hidden Markov models
• Debiasing Stochastic Gradient Descent to handle missing values
• An efficient particle-based online EM algorithm for general state-space models
• An Empirical Study on Academic Commentary and Its Implications on Reading and Writing
• Stochastic Discriminative EM
• Stochastic Collapsed Variational Bayesian Inference for Latent Dirichlet Allocation
• Inference Graphs for CNN Interpretation
• Parallel Stochastic Optimization Framework for Large-Scale Non-Convex Stochastic Problems
• Distributed localization of a RF target in NLOS environments
• EMFlow: Data Imputation in Latent Space via EM and Deep Flow Models
• A factor graph EM algorithm for inference of kinetic microstates from patch clamp measurements
• Unbiased Pairwise Learning to Rank in Recommender Systems
• Mini-batch learning of exponential family finite mixture models
• Geom-SPIDER-EM: Faster Variance Reduced Stochastic Expectation Maximization for Nonconvex Finite-Sum Optimization
• Efficient convergence through adaptive learning in sequential Monte Carlo Expectation Maximization
• Stochastic Multidimensional Scaling
• Fast Incremental Expectation Maximization for finite-sum optimization: nonasymptotic convergence
• Probabilistic sequential matrix factorization
• Stochastic EM methods with Variance Reduction for Penalised PET Reconstructions
• A Novel Document Generation Process for Topic Detection based on Hierarchical Latent Tree Models
• On PyTorch Implementation of Density Estimators for von Mises-Fisher and Its Mixture
• Online EM for Functional Data
• Homeomorphic-Invariance of EM: Non-Asymptotic Convergence in KL Divergence for Exponential Families via Mirror Descent