• Home
  • Search
  • An alternative to EM for Gaussian mixture models: batch and stochastic Riemannian optimization
  • Cite Icon55
  • https://doi.org/10.1007/s10107-019-01381-4Copy DOI Icon

An alternative to EM for Gaussian mixture models: batch and stochastic Riemannian optimization

Show More
  • Abstract
  • Literature Map
  • References
  • Citations
  • Similar Papers
Abstract

We consider maximum likelihood estimation for Gaussian Mixture Models (Gmm s). This task is almost invariably solved (in theory and practice) via the Expectation Maximization (EM) algorithm. EM owes its success to various factors, of which is its ability to fulfill positive definiteness constraints in closed form is of key importance. We propose an alternative to EM grounded in the Riemannian geometry of positive definite matrices, using which we cast Gmm parameter estimation as a Riemannian optimization problem. Surprisingly, such an out-of-the-box Riemannian formulation completely fails and proves much inferior to EM. This motivates us to take a closer look at the problem geometry, and derive a better formulation that is much more amenable to Riemannian optimization. We then develop Riemannian batch and stochastic gradient algorithms that outperform EM, often substantially. We provide a non-asymptotic convergence analysis for our stochastic method, which is also the first (to our knowledge) such global analysis for Riemannian stochastic gradient. Numerous empirical results are included to demonstrate the effectiveness of our methods.

Similar Papers
  • Research Article
  • Citations31

Biased stochastic conjugate gradient algorithm with adaptive step size for nonconvex problems

  • Sep 22, 2023
  • Expert Systems with Applications
  • Ruping Huang +3
  • Research Article
  • Citations67

Dualistic differential geometry of positive definite matrices and its applications to related problems

  • Nov 01, 1996
  • Linear Algebra and its Applications
  • Atsumi Ohara +2
  • Conference Article
  • Citations7

A Novel Constrained Adaptive Algorithm Using the Conjugate Gradient Method for Smart Antennas

  • Nov 01, 2007
  • Lei Wang +1
  • Research Article
  • Citations3

A comparison of stochastic gradient and minimum entropy deconvolution algorithms

  • Sep 01, 1988
  • Signal Processing
  • A.T Walden
  • Research Article
  • Citations15

A linearly convergent stochastic recursive gradient method for convex optimization

  • Feb 17, 2020
  • Optimization Letters
  • Yan Liu +2
  • PDF
  • Research Article
  • Citations3

Joint Multitarget Detection and Tracking in Multipath Environment Using Expectation Maximization Algorithm

  • Jan 01, 2021
  • IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
  • Zhihua Li +4
  • Conference Article

Low complexity variable step size mechanisms for constrained minimum variance CDMA receivers

  • May 17, 2004
  • R.C De Lamare +1
  • Conference Article

Several Stochastic Gradient Algorithms for Nonlinear Systems with Hard Nonlinearities

  • Aug 01, 2015
  • Jia Tang
  • Research Article
  • Citations57

An EM algorithm for mapping binary disease loci: application to fibrosarcoma in a four-way cross mouse family.

  • Apr 01, 2003
  • Genetical research
  • Shizhong Xu +4
  • Research Article
  • Citations8

Stochastic gradient method with accelerated stochastic dynamics

  • Mar 01, 2016
  • Journal of Physics: Conference Series
  • Masayuki Ohzeki
  • Book Chapter
  • Citations1

Using a Genetic Algorithm for Selection of Starting Conditions for the EM Algorithm for Gaussian Mixture Models

  • Jan 01, 2016
  • Wojciech Kwedlo
  • Conference Article

Enhanced Mixture Population Monte Carlo Via Stochastic Optimization and Markov Chain Monte Carlo Sampling

  • May 01, 2020
  • Yousef El-Laham +2
  • Conference Article

The probability distribution of parameters learned with the EM algorithm

  • Jun 03, 1996
  • K Ikeda +1
  • Conference Article
  • Citations2

Using unsupervised learning to determine risk level for left ventricular diastolic dysfunction

  • Nov 01, 2014
  • Kaidi Ma +3
  • Research Article
  • Citations11

Simulation study on missing data imputation methods for longitudinal data in cohort studies

  • Oct 10, 2021
  • Zhonghua liu xing bing xue za zhi = Zhonghua liuxingbingxue zazhi
  • Y M Li +5
Cactus Communications logo

Copyright 2026 Cactus Communications. All rights reserved.