Interpolated Adversarial Training: Achieving Robust Neural Networks without Sacrificing Accuracy.
Alex LambVikas VermaJuho KannalaYoshua BengioPublished in: CoRR (2019)
Keyphrases
- neural network
- training process
- highly accurate
- training algorithm
- artificial neural networks
- pattern recognition
- neural network training
- self organizing maps
- multi layer perceptron
- neural nets
- computational cost
- high accuracy
- registration errors
- feedforward neural networks
- feed forward neural networks
- genetic algorithm
- neural network model
- classification accuracy
- prediction accuracy
- test set
- feature selection
- supervised learning
- training examples
- decision trees
- sufficiently accurate
- classification performances
- training speed
- recurrent networks
- error rate
- multi agent
- recurrent neural networks
- computational efficiency
- data sets
- fault diagnosis