A predictive deviance criterion for selecting a generative model in semi-supervised classification.
Vincent VandewalleChristophe BiernackiGilles CeleuxGérard GovaertPublished in: Comput. Stat. Data Anal. (2013)
Keyphrases
- generative model
- semi supervised classification
- semi supervised
- semi supervised learning
- labeled data
- unlabeled data
- parameter learning
- probabilistic model
- bayesian framework
- unsupervised learning
- pairwise constraints
- em algorithm
- supervised learning
- pairwise
- active learning
- conditional random fields
- topic models
- hidden variables
- co training
- labeled and unlabeled data
- metric learning
- data sets
- training data