Serf: Towards better training of deep neural networks using log-Softplus ERror activation Function.
Sayan NagMayukh BhattacharyyaAnuraag MukherjeeRohit KunduPublished in: WACV (2023)
Keyphrases
- activation function
- neural network
- feed forward neural networks
- backpropagation algorithm
- error function
- feedforward neural networks
- single hidden layer
- back propagation
- hidden layer
- multi layer perceptron
- training algorithm
- feed forward
- artificial neural networks
- training phase
- multilayer perceptron
- hidden neurons
- neural nets
- training process
- radial basis function
- learning rate
- extreme learning machine
- network architecture
- fuzzy neural network
- basis functions
- recurrent neural networks
- multi layer
- hidden nodes
- pattern recognition
- output layer
- single layer
- rbf neural network
- supervised learning
- expert systems
- training speed
- neural network model