MITA: Multi-Input Adaptive Activation Function for Accurate Binary Neural Network Hardware.
Peiqi ZhangShinya Takamaeda-YamazakiPublished in: IEICE Trans. Inf. Syst. (2023)
Keyphrases
- pattern recognition
- neural network
- activation function
- single hidden layer
- hidden units
- feedforward neural networks
- hidden layer
- feed forward neural networks
- extreme learning machine
- back propagation
- artificial neural networks
- hidden neurons
- connection weights
- multilayer perceptron
- output layer
- nonlinear functions
- feed forward
- backpropagation algorithm
- radial basis function
- neural network model
- network architecture
- sigmoid function
- hardware implementation
- genetic algorithm
- neural nets
- basis functions
- learning rate
- linear combination
- hidden nodes
- input output
- support vector regression
- multi layer perceptron
- bp neural network
- fuzzy neural network
- rbf network
- fault diagnosis
- non stationary
- single layer
- input patterns
- training algorithm
- control algorithm
- particle swarm optimization
- input data
- expert systems
- decision making
- feature selection