A method for training recurrent neural networks for classification by building basins of attraction.
Roelof K. BrouwerPublished in: Neural Networks (1995)
Keyphrases
- experimental evaluation
- significant improvement
- high precision
- high accuracy
- synthetic data
- data sets
- training process
- pairwise
- classification accuracy
- computationally efficient
- fully automatic
- test data
- training phase
- recurrent neural networks
- feed forward
- optimization method
- detection method
- training samples
- cost function
- preprocessing
- objective function
- feature extraction
- image segmentation