Morfessor EM+Prune: Improved Subword Segmentation with Expectation Maximization and Pruning.
Stig-Arne GrönroosSami VirpiojaMikko KurimoPublished in: LREC (2020)
Keyphrases
- expectation maximization
- em algorithm
- image segmentation
- search space
- expectation maximisation
- mixture model
- maximum likelihood
- segmentation algorithm
- unsupervised learning
- generative model
- gaussian mixture model
- parameter estimation
- level set
- probabilistic model
- maximum likelihood estimation
- shape prior
- medical images
- gaussian mixture
- bayesian framework
- pruning strategy
- n gram
- object segmentation
- multiscale
- region growing
- gaussian distribution
- segmentation accuracy
- likelihood function
- mixture of gaussians
- probability density function
- finite mixture model
- energy function
- k means
- image analysis
- maximum a posteriori
- density estimation
- background subtraction
- active contours
- segmented images
- graph cuts
- log likelihood function
- gradient ascent
- bayesian information criterion
- image regions
- segmentation method
- optimal solution