Overcoming Mean-Field Approximations in Recurrent Gaussian Process Models.
Alessandro Davide IalongoMark van der WilkJames HensmanCarl Edward RasmussenPublished in: CoRR (2019)
Keyphrases
- gaussian process models
- gaussian processes
- gaussian process
- free energy
- closed form
- variational methods
- human pose estimation
- variational approximation
- markov random field
- em algorithm
- hyperparameters
- bayesian inference
- bayesian framework
- markov networks
- particle filter
- surveillance system
- learning algorithm
- belief propagation
- regression model
- model selection
- expectation maximization
- image sequences
- three dimensional
- computer vision