Geom-Spider-EM: Faster Variance Reduced Stochastic Expectation Maximization for Nonconvex Finite-Sum Optimization.
Gersende FortEric MoulinesHoi-To WaiPublished in: ICASSP (2021)
Keyphrases
- expectation maximization
- em algorithm
- global optimization
- probabilistic model
- mixture model
- optimization problems
- maximum likelihood
- parameter estimation
- generative model
- stochastic optimization
- gaussian mixture model
- objective function
- stochastic search
- maximum a posteriori
- bayesian framework
- nonlinear programming
- optimization algorithm
- stochastic programming
- image segmentation
- unsupervised learning
- finite mixture model
- gaussian mixture
- maximum likelihood estimation
- likelihood function
- monte carlo
- discrete random variables
- probability density function
- gaussian distribution
- mathematical program
- optimal control problems
- mixture of gaussians
- constrained optimization
- density estimation
- covariance matrix
- gaussian densities
- image restoration and reconstruction
- optimization method
- web mining
- bayesian networks