Linear Time Model Selection for Mixture of Heterogeneous Components.
Ryohei FujimakiSatoshi MorinagaMichinari MommaKenji AokiTakayuki NakataPublished in: ACML (2009)
Keyphrases
- model selection
- mixture model
- cross validation
- hyperparameters
- bayesian learning
- machine learning
- statistical learning
- parameter estimation
- regression model
- generalization error
- sample size
- information criterion
- feature selection
- meta learning
- variable selection
- model selection criteria
- statistical inference
- motion segmentation
- error estimation
- unsupervised learning
- bayesian methods
- gaussian mixture model
- generalization bounds
- training data
- expectation maximization
- gaussian process
- worst case
- support vector machine
- likelihood function
- selection criterion
- bayesian information criterion
- leave one out cross validation