Model selection via marginal likelihood estimation by combining thermodynamic integration and gradient matching.
Benn MacdonaldDirk HusmeierPublished in: Stat. Comput. (2019)
Keyphrases
- marginal likelihood
- model selection
- parameter estimation
- hyperparameters
- cross validation
- selection criterion
- model selection criteria
- information criterion
- regression model
- bayesian learning
- gaussian process
- sample size
- bayesian information criterion
- feature selection
- free energy
- approximate inference
- mixture model
- edge detection
- variable selection
- closed form
- maximum likelihood
- leave one out cross validation
- machine learning
- density estimation
- feature vectors