Efficient Implementation of Activation Functions for LSTM accelerators.
Yi Sheng ChongWang Ling GohYew Soon OngVishnu P. NambiarAnh Tuan DoPublished in: VLSI-SoC (2021)
Keyphrases
- efficient implementation
- activation function
- recurrent neural networks
- neural network
- hidden layer
- hardware implementation
- feed forward
- artificial neural networks
- graphics processing units
- field programmable gate array
- back propagation
- neural nets
- neural architecture
- multilayer perceptron
- fuzzy neural network
- hidden nodes
- active set
- learning rate
- computing systems
- network architecture
- training phase
- basis functions
- multi layer perceptron
- single chip
- radial basis function
- reinforcement learning
- convergence rate