Convergence suppression and divergence facilitation: new approach to prune hidden layer and weights of feedforward neural networks
説明
A pruning algorithm is devised for multilayer multi-output feedforward perceptron networks. The algorithm efficiently reduces the total number of hidden units and the number of weights in the output layer. Test examples include network pruning for the IRIS classifier and for the bitmap digit classifier.
収録刊行物
-
- Proceedings of ISCAS'95 - International Symposium on Circuits and Systems
-
Proceedings of ISCAS'95 - International Symposium on Circuits and Systems 1 121-124, 2002-11-19
IEEE