Active Subsampling Using Deep Generative Models by Maximizing Expected Information Gain.
Koen C. E. van de CampHamdi JoudehDuarte J. AntunesRuud J. G. van SlounPublished in: ICASSP (2023)
Keyphrases
- information gain
- generative model
- probabilistic model
- feature selection
- decision trees
- text categorization
- chi squared
- discriminative learning
- deep belief networks
- mutual information
- discriminative models
- prior knowledge
- semi supervised
- latent dirichlet allocation
- em algorithm
- naive bayes
- topic models
- feature subset
- data sets
- active learning
- object recognition
- natural language
- information retrieval