Towards Developing Probabilistic Generative Models for Reasoning with Natural Language Representations.
Daniel MarcuAna-Maria PopescuPublished in: CICLing (2005)
Keyphrases
- generative model
- natural language
- probabilistic model
- feature representations
- knowledge representation
- posterior probability
- mixture model
- prior knowledge
- object categories
- deep belief networks
- conditional random fields
- em algorithm
- natural language processing
- visual patterns
- discriminative models
- maximum entropy principle
- discriminative learning
- hierarchical models
- representational power
- semi supervised
- higher level
- hidden variables
- generative and discriminative models
- hierarchical hidden markov models