Overcoming the local-minimum problem in training multilayer perceptrons by gradual deconvexification.
James Ting-Ho LoYichuan GuiYun PengPublished in: IJCNN (2013)
Keyphrases
- multilayer perceptron
- backpropagation algorithm
- radial basis function network
- feedforward neural networks
- back propagation
- training algorithm
- artificial neural networks
- neural network
- multilayer neural network
- radial basis function
- hidden layer
- neural network model
- commonly applied
- training process
- training phase
- number of hidden neurons
- belief networks
- training set
- feed forward neural networks
- activation function
- multi layer perceptron
- bp algorithm
- learning speed
- neural architecture
- basis functions
- expert systems