Multi-objectivization and Surrogate Modelling for Neural Network Hyper-parameters Tuning.
Martin PilátRoman NerudaPublished in: ICIC (3) (2013)
Keyphrases
- hyperparameters
- neural network
- model selection
- cross validation
- closed form
- bayesian framework
- support vector
- bayesian inference
- noise level
- parameter settings
- gaussian process
- random sampling
- prior information
- maximum likelihood
- em algorithm
- variational bayes
- incremental learning
- maximum a posteriori
- sample size
- artificial neural networks
- incomplete data
- posterior distribution
- bayesian methods
- missing values
- data mining
- parameter values
- np hard
- feature selection