A fast and efficient pre-training method based on layer-by-layer maximum discrimination for deep neural networks.
Seyyede Zohreh SeyyedsalehiSeyyed Ali SeyyedsalehiPublished in: Neurocomputing (2015)
Keyphrases
- neural network
- training process
- multi layer
- training algorithm
- preprocessing
- high efficiency
- high accuracy
- computational cost
- highly efficient
- high precision
- dynamic programming
- significant improvement
- clustering method
- segmentation method
- computational complexity
- objective function
- feedforward neural networks
- rbf network
- feed forward neural networks
- model selection
- cost function
- pairwise
- pattern recognition