Sinkhorn EM: An Expectation-Maximization algorithm based on entropic optimal transport.
Gonzalo E. MenaAmin NejatbakhshErdem VarolJonathan Niles-WeedPublished in: CoRR (2020)
Keyphrases
- expectation maximization
- em algorithm
- dynamic programming
- k means
- optimal solution
- learning algorithm
- maximum likelihood estimation
- times faster
- maximum likelihood
- probabilistic model
- significant improvement
- experimental evaluation
- cost function
- objective function
- globally optimal
- recognition algorithm
- mixture model
- gradient projection
- computational cost
- image processing
- optimal parameters
- gaussian mixture
- path planning
- optimization algorithm
- detection algorithm
- clustering method
- gaussian mixture model
- parameter estimation
- data structure
- worst case
- high accuracy
- neural network
- preprocessing
- generative model
- locally optimal
- optimality criterion
- log likelihood function
- search algorithm