SAU: Smooth activation function using convolution with approximate identities.
Koushik BiswasSandeep KumarShilpak BanerjeeAshish Kumar PandeyPublished in: CoRR (2021)
Keyphrases
- activation function
- neural network
- artificial neural networks
- feed forward
- feed forward neural networks
- back propagation
- hidden layer
- multilayer perceptron
- feedforward neural networks
- neural nets
- chaotic neural network
- learning rate
- network size
- basis functions
- hidden nodes
- network architecture
- radial basis function
- image processing
- extreme learning machine
- autoregressive
- recurrent neural networks