Unreasonable Effectiveness of Last Hidden Layer Activations for Adversarial Robustness.
Ömer Faruk TunaFerhat Özgür ÇatakM. Taner EskilPublished in: COMPSAC (2022)
Keyphrases
- hidden layer
- feed forward
- back propagation
- feedforward neural networks
- activation function
- neural network
- radial basis function
- recurrent neural networks
- learning rate
- artificial neural networks
- feed forward neural networks
- multilayer perceptron
- neural nets
- output layer
- radial basis function neural network
- rbf neural network
- convergence rate
- input output
- nearest neighbor
- small number
- image segmentation
- hidden nodes
- knowledge base