Boosting neural network feature extraction by reduced accuracy activation functions.
Aistis RaudysPublished in: Pattern Recognit. (2003)
Keyphrases
- activation function
- neural network
- feature extraction
- neural architecture
- feed forward neural networks
- back propagation
- hidden layer
- artificial neural networks
- neural nets
- feed forward
- feature selection
- connection weights
- feedforward neural networks
- multilayer perceptron
- pattern recognition
- network architecture
- learning rate
- hidden nodes
- hidden neurons
- sigmoid function
- multi layer perceptron
- genetic algorithm
- bp neural network
- radial basis function
- support vector machine svm
- classification accuracy
- face recognition
- image processing
- multi layer
- neural network model
- fault diagnosis
- fuzzy neural network
- dimensionality reduction
- fuzzy logic
- feature vectors
- high dimensional
- learning algorithm