Skip to Main content Skip to Navigation
New interface
Conference papers

Geom-SPIDER-EM: Faster Variance Reduced Stochastic Expectation Maximization for Nonconvex Finite-Sum Optimization

Abstract : The Expectation Maximization (EM) algorithm is a key reference for inference in latent variable models; unfortunately, its computational cost is prohibitive in the large scale learning setting. In this paper, we propose an extension of the Stochastic Path-Integrated Differential EstimatoR EM (SPIDER-EM) and derive complexity bounds for this novel algorithm, designed to solve smooth nonconvex finite-sum optimization problems. We show that it reaches the same state of the art complexity bounds as SPIDER-EM; and provide conditions for a linear rate of convergence. Numerical results support our findings.
Complete list of metadata

https://hal.archives-ouvertes.fr/hal-03021394
Contributor : Gersende Fort Connect in order to contact the contributor
Submitted on : Monday, February 8, 2021 - 8:53:57 PM
Last modification on : Tuesday, October 25, 2022 - 11:58:11 AM

File

GeomSPIDEREM_HAL.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-03021394, version 2
  • ARXIV : 2011.12392

Citation

Gersende Fort, Eric Moulines, Hoi-To Wai. Geom-SPIDER-EM: Faster Variance Reduced Stochastic Expectation Maximization for Nonconvex Finite-Sum Optimization. 2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), Jun 2021, Toronto, Canada. ⟨hal-03021394v2⟩

Share

Metrics

Record views

159

Files downloads

102