A New Proposition of the Activation Function for Significant Improvement of Neural Networks Performance.
Jaroslaw BilskiAlexander I. GalushkinPublished in: ICAISC (1) (2016)
Keyphrases
- activation function
- neural network
- significant improvement
- back propagation
- artificial neural networks
- hidden layer
- feed forward neural networks
- feed forward
- neural nets
- multilayer perceptron
- hidden neurons
- single hidden layer
- feedforward neural networks
- backpropagation algorithm
- learning rate
- hidden nodes
- fuzzy neural network
- pattern recognition
- network architecture
- radial basis function
- training phase
- extreme learning machine
- sigmoid function
- fuzzy logic
- input space
- single layer
- recurrent neural networks
- neural network model
- support vector machine svm
- chaotic neural network
- decision making
- training algorithm
- hidden units
- support vector machine