On the estimation of hyperparameters for Bayesian system identification with exponentially decaying kernels.
Ruggero CarliTianshi ChenAlessandro ChiusoLennart LjungGianluigi PillonettoPublished in: CDC (2012)
Keyphrases
- hyperparameters
- gaussian processes
- posterior distribution
- bayesian inference
- support vector
- model selection
- maximum likelihood
- bayesian methods
- cross validation
- gaussian process
- closed form
- bayesian framework
- covariance function
- em algorithm
- parameter estimation
- random sampling
- conjugate priors
- marginal likelihood
- noise level
- prior information
- maximum a posteriori
- sample size
- incremental learning
- regularization parameter
- posterior probability
- probabilistic model
- incomplete data
- gaussian process regression
- expectation maximization
- kernel function
- parameter space
- probability distribution
- bayesian networks
- graph cuts
- markov random field
- missing values
- machine learning