The Effects of Hyperparameters on SGD Training of Neural Networks.
Thomas M. BreuelPublished in: CoRR (2015)
Keyphrases
- hyperparameters
- neural network
- model selection
- stochastic gradient descent
- cross validation
- training process
- regularization parameter
- bayesian inference
- bayesian framework
- grid search
- random sampling
- em algorithm
- training set
- closed form
- gaussian process
- prior information
- noise level
- support vector
- maximum a posteriori
- sample size
- maximum likelihood
- gaussian processes
- incremental learning
- back propagation
- incomplete data
- artificial neural networks
- supervised learning
- upper bound
- multiscale
- genetic algorithm
- missing values