Stochastic resonance induced continuous activation functions in a neural network consisting of threshold elements.
Shuhei IkemotoFabio Dalla LiberaKoh HosodaPublished in: IJCNN (2016)
Keyphrases
- activation function
- stochastic resonance
- neural network
- continuous functions
- image enhancement
- contrast enhancement
- signal to noise ratio
- artificial neural networks
- feed forward neural networks
- back propagation
- hidden layer
- feed forward
- sigmoid function
- neural nets
- neural architecture
- connection weights
- additive noise
- multilayer perceptron
- low signal to noise ratio
- learning rate
- signal processing
- network architecture
- basis functions
- multi layer perceptron
- fuzzy neural network
- neural network model
- genetic algorithm
- pattern recognition
- input space
- fuzzy logic
- recurrent neural networks
- maximum likelihood
- training phase
- fault diagnosis
- edge detection
- wavelet transform
- visual quality
- image processing
- learning algorithm
- human visual system