Do Transformers Dream of Inference, or Can Pretrained Generative Models Learn Implicit Inferential Rules?
Zhengzhong LiangMihai SurdeanuPublished in: Insights (2020)
Keyphrases
- generative model
- boltzmann machine
- learning rules
- parameter learning
- probabilistic model
- discriminative learning
- discriminative models
- prior knowledge
- mixture model
- representational power
- efficient learning
- conditional random fields
- em algorithm
- latent dirichlet allocation
- maximum entropy principle
- topic models
- generative and discriminative models
- probabilistic inference
- pitman yor process
- hierarchical hidden markov models
- bayesian inference
- bayesian networks
- naive bayes models
- semi supervised
- similarity measure
- object categories
- information extraction