Bounding the Test Log-Likelihood of Generative Models.
Yoshua BengioLi YaoPublished in: ICLR (Poster) (2014)
Keyphrases
- generative model
- log likelihood
- em algorithm
- mixture model
- density estimation
- expectation maximization
- probabilistic model
- maximum likelihood
- bayesian framework
- information theoretic
- discriminative learning
- hierarchical hidden markov models
- posterior probability
- hidden variables
- conditional random fields
- upper bound
- prior knowledge
- exponential family
- discriminative models
- gaussian mixture model
- parameter estimation
- closed form
- probability density function
- semi supervised
- maximum likelihood estimation
- scoring function
- generative and discriminative models
- pairwise
- search engine