ProAct: Progressive Training for Hybrid Clipped Activation Function to Enhance Resilience of DNNs.
Seyedhamidreza MousaviMohammad Hasan AhmadilivaniJaan RaikMaksim JenihhinMasoud DaneshtalabPublished in: CoRR (2024)
Keyphrases
- activation function
- feed forward neural networks
- feedforward neural networks
- neural network
- backpropagation algorithm
- learning speed
- single hidden layer
- training phase
- hidden layer
- artificial neural networks
- back propagation
- feed forward
- learning rate
- extreme learning machine
- training set
- training algorithm
- training examples
- network architecture
- basis functions
- training process
- chaotic neural network
- multi layer
- multi layer perceptron
- neural nets
- radial basis function
- hidden nodes
- multilayer perceptron
- supervised learning
- output layer
- sigmoid function
- training speed
- fuzzy neural network
- prior knowledge
- learning algorithm