Wasserstein proximal operators describe score-based generative models and resolve memorization.
Benjamin J. ZhangSiting LiuWuchen LiMarkos A. KatsoulakisStanley J. OsherPublished in: CoRR (2024)
Keyphrases
- generative model
- probabilistic model
- mixture model
- prior knowledge
- hierarchical hidden markov models
- discriminative models
- discriminative learning
- topic models
- semi supervised
- em algorithm
- maximum entropy principle
- generative and discriminative models
- expectation maximization
- object categories
- hidden variables
- conditional random fields
- natural images
- higher order
- representational power
- discriminatively trained
- object recognition