Approximate Inference Turns Deep Networks into Gaussian Processes.
Mohammad Emtiyaz KhanAlexander ImmerEhsan AbediMaciej KorzepaPublished in: NeurIPS (2019)
Keyphrases
- approximate inference
- gaussian process
- gaussian processes
- expectation propagation
- regression model
- latent variables
- hyperparameters
- bayesian framework
- exact inference
- model selection
- semi supervised
- belief propagation
- dynamic bayesian networks
- message passing
- graphical models
- human pose estimation
- parameter estimation
- probabilistic inference
- bayesian methods
- conditional random fields
- maximum a posteriori
- reinforcement learning
- em algorithm
- co occurrence