Training Deep Neural Networks by optimizing over nonlocal paths in hyperparameter space.
Vlad PushkarovJonathan EfroniMykola MaksymenkoMaciej Koch-JanuszPublished in: CoRR (2019)
Keyphrases
- neural network
- training process
- training algorithm
- feedforward neural networks
- pattern recognition
- deep architectures
- feed forward neural networks
- space time
- machine learning
- neural network training
- neural network model
- back propagation
- low dimensional
- fuzzy logic
- shortest path
- image restoration
- search space
- feed forward
- artificial neural networks
- genetic algorithm
- image segmentation
- multi layer perceptron
- error back propagation
- training samples
- backpropagation algorithm
- video sequences
- feature space
- high dimensional
- gaussian process
- multilayer perceptron