Efficient decomposition of latent representation in generative models.
Vsevolod NikulinJun TaniPublished in: SSCI (2020)
Keyphrases
- generative model
- latent space
- prior knowledge
- representational power
- fisher kernel
- mixture model
- probabilistic model
- discriminative models
- discriminative learning
- generative process
- image representation
- visual patterns
- latent dirichlet allocation
- latent variables
- em algorithm
- discriminative learning framework
- maximum entropy principle
- object categories
- hierarchical hidden markov models
- machine learning
- object detection
- semi supervised
- computer vision