Flex-SFU: Accelerating DNN Activation Functions by Non-Uniform Piecewise Approximation.
Enrico ReggianiRenzo AndriLukas CavigelliPublished in: DAC (2023)
Keyphrases
- activation function
- continuous functions
- neural network
- feed forward
- neural architecture
- artificial neural networks
- hidden layer
- back propagation
- learning rate
- neural nets
- basis functions
- hidden nodes
- radial basis function
- multilayer perceptron
- network architecture
- artificial intelligence
- training process
- support vector machine svm
- fuzzy neural network
- multi layer perceptron
- support vector machine
- feature selection