Maximum of Marginal Likelihood Criterion instead of Cross-Validation for Designing of Artificial Neural Networks.
Zenon WaszczyszynMarek SlonskiPublished in: ICAISC (2008)
Keyphrases
- marginal likelihood
- cross validation
- model selection
- artificial neural networks
- information criterion
- hyperparameters
- selection criterion
- gaussian process
- generalization error
- variable selection
- neural network
- sample size
- support vector
- bayesian information criterion
- leave one out cross validation
- regression model
- parameter estimation
- machine learning
- training set
- closed form
- bayesian methods
- probability model
- exponential family
- bayesian networks