Out-distribution training confers robustness to deep neural networks.
Mahdieh AbbasiChristian GagnéPublished in: CoRR (2018)
Keyphrases
- neural network
- training process
- training algorithm
- feedforward neural networks
- feed forward neural networks
- back propagation
- pattern recognition
- training set
- backpropagation algorithm
- multi layer perceptron
- test set
- deep architectures
- artificial neural networks
- uniformly distributed
- genetic algorithm
- neural network training
- training and test data
- training patterns
- error back propagation
- fuzzy logic
- artificial intelligence
- data distribution
- radial basis function network
- self organizing maps
- deep learning
- expert systems
- neural network structure
- probability distribution
- training examples
- input pattern
- gaussian distribution
- computational efficiency
- probability density function
- feed forward
- training phase
- multilayer perceptron