Parameterizing Activation Functions for Adversarial Robustness.
Sihui DaiSaeed MahloujifarPrateek MittalPublished in: CoRR (2021)
Keyphrases
- activation function
- neural network
- complex valued
- feed forward
- artificial neural networks
- feed forward neural networks
- neural architecture
- back propagation
- hidden layer
- learning rate
- neural nets
- network architecture
- multilayer perceptron
- training phase
- hidden nodes
- basis functions
- recurrent neural networks
- multi layer perceptron
- fuzzy sets
- feature selection
- machine learning