Loss function based second-order Jensen inequality and its application to particle variational inference.
Futoshi FutamiTomoharu IwataNaonori UedaIssei SatoMasashi SugiyamaPublished in: NeurIPS (2021)
Keyphrases
- loss function
- variational inference
- bayesian inference
- gaussian process
- topic models
- probabilistic graphical models
- probabilistic model
- variational methods
- pairwise
- posterior distribution
- mixture model
- latent dirichlet allocation
- closed form
- higher order
- exponential family
- support vector
- graphical models
- approximate inference
- regularization term
- exact inference
- bayesian framework
- hyperparameters
- expectation maximization
- latent variables
- gaussian mixture model
- markov random field
- semi supervised
- probability distribution
- optical flow
- graph cuts