Properties of the Stochastic Approximation EM Algorithm with Mini-batch Sampling

Abstract : To speed up convergence a mini-batch version of the Monte Carlo Markov Chain Stochas-tic Approximation Expectation-Maximization (MCMC-SAEM) algorithm for general latent variable models is proposed. For exponential models the algorithm is shown to be conver-gent under classical conditions as the number of iterations increases. Numerical experiments illustrate the performance of the mini-batch algorithm in various models. In particular, we highlight that an appropriate choice of the mini-batch size results in a tremendous speed-up of the convergence of the sequence of estimators generated by the algorithm. Moreover, insights on the effect of the mini-batch size on the limit distribution are presented.
Document type :
Preprints, Working Papers, ...
Complete list of metadatas

https://hal.archives-ouvertes.fr/hal-02189215
Contributor : Catherine Matias <>
Submitted on : Friday, July 19, 2019 - 11:15:17 AM
Last modification on : Friday, September 20, 2019 - 3:26:02 PM

Files

preprint.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-02189215, version 1
  • ARXIV : 1907.09164

Citation

Estelle Kuhn, Catherine Matias, Tabea Rebafka. Properties of the Stochastic Approximation EM Algorithm with Mini-batch Sampling. 2019. ⟨hal-02189215⟩

Share

Metrics

Record views

45

Files downloads

45