Perturbative corrections for approximate inference in Gaussian latent variable models.
Manfred OpperUlrich PaquetOle WintherPublished in: J. Mach. Learn. Res. (2013)
Keyphrases
- approximate inference
- latent variable models
- latent variables
- expectation propagation
- gaussian process
- probabilistic model
- exact inference
- variational methods
- posterior distribution
- prior knowledge
- random variables
- graphical models
- topic models
- belief propagation
- marginal likelihood
- probabilistic inference
- hidden variables
- maximum likelihood
- structured prediction
- machine learning
- message passing
- free energy
- markov random field
- parameter estimation
- markov chain monte carlo
- least squares
- hidden markov models
- gaussian processes
- bayesian inference
- active learning
- gaussian distribution
- training set
- generative model
- bayesian networks
- feature selection
- stereo matching