A Divergence Bound for Hybrids of MCMC and Variational Inference and an Application to Langevin Dynamics and SGVI.
Justin DomkePublished in: ICML (2017)
Keyphrases
- variational inference
- posterior distribution
- bayesian inference
- markov chain monte carlo
- dirichlet process
- latent variables
- approximate inference
- probability distribution
- topic models
- probabilistic model
- latent dirichlet allocation
- bayesian framework
- variational methods
- hyperparameters
- parameter estimation
- lower bound
- mixture model
- upper bound
- gaussian process
- maximum a posteriori
- generative model
- graphical models
- exponential family
- hidden variables
- probabilistic graphical models
- markov chain
- posterior probability
- probabilistic inference
- belief propagation
- message passing
- prior information
- expectation maximization
- level set
- least squares
- gaussian distribution
- support vector