Boosting Black-Box Variational Inference by Incorporating the Natural Gradient.
Felix TrusheimAlexandru ConduracheAlfred MertinsPublished in: ICPR (2018)
Keyphrases
- black box
- variational inference
- natural gradient
- bayesian inference
- topic models
- learning rate
- probabilistic model
- independent component analysis
- posterior distribution
- latent dirichlet allocation
- variational methods
- blind source separation
- gaussian process
- mixture model
- closed form
- learning algorithm
- latent variables
- test cases
- generative model
- approximate inference
- optic flow
- feature selection
- bayesian framework
- prior information
- maximum a posteriori
- markov random field
- bayesian networks
- data sets