A Test of Relative Similarity For Model Selection in Generative Models.
Wacha BounliphoneEugene BelilovskyMatthew B. BlaschkoIoannis AntonoglouArthur GrettonPublished in: ICLR (Poster) (2016)
Keyphrases
- model selection
- generative model
- mixture model
- cross validation
- likelihood function
- hyperparameters
- probabilistic model
- parameter estimation
- discriminative learning
- selection criterion
- model selection criteria
- statistical learning
- bayesian framework
- hidden variables
- discriminative models
- machine learning
- hierarchical hidden markov models
- regression model
- unsupervised learning
- similarity measure
- sample size
- posterior probability
- generalization error
- gaussian process
- prior knowledge
- gaussian mixture model
- em algorithm
- expectation maximization
- feature selection
- conditional random fields
- topic models
- probability density function
- information retrieval