Variational inference via Wasserstein gradient flows.
Marc LambertSinho ChewiFrancis R. BachSilvère BonnabelPhilippe RigolletPublished in: NeurIPS (2022)
Keyphrases
- variational inference
- bayesian inference
- topic models
- posterior distribution
- gaussian process
- probabilistic model
- mixture model
- variational methods
- latent dirichlet allocation
- closed form
- active contour model
- exponential family
- approximate inference
- curve evolution
- active contours
- probability distribution
- latent variables
- expectation maximization
- hyperparameters
- bayesian framework
- graphical models
- prior information
- model selection
- parameter estimation
- density estimation
- belief propagation
- image reconstruction
- log likelihood
- pairwise