The SVHN Dataset Is Deceptive for Probabilistic Generative Models Due to a Distribution Mismatch.
Tim Z. XiaoJohannes ZennRobert BamlerPublished in: CoRR (2023)
Keyphrases
- generative model
- probabilistic model
- mixture model
- posterior probability
- bayesian framework
- semi supervised
- discriminative models
- em algorithm
- prior knowledge
- maximum entropy principle
- conditional random fields
- latent dirichlet allocation
- generative and discriminative models
- random variables
- representational power
- topic models
- graphical models
- discriminative learning
- probability distribution
- expectation maximization
- language model
- hierarchical hidden markov models