Probabilistic relabelling strategies for the label switching problem in Bayesian mixture models.
Matthew SperrinThomas JakiErnst WitPublished in: Stat. Comput. (2010)
Keyphrases
- active learning
- mixture model
- generalized em algorithm
- generative model
- probabilistic model
- posterior probability
- maximum likelihood
- em algorithm
- bayesian networks
- gaussian mixture model
- dirichlet prior
- density estimation
- language model
- bayesian inference
- expectation maximization
- model selection
- probability density function
- unsupervised learning
- automatic model selection
- finite mixture models
- bayesian framework
- variational inference
- mixture modeling
- posterior distribution
- prior distribution
- object recognition
- conditional probabilities
- real time