A constructive algorithm for feedforward neural networks
Jinhua Xu, Daniel W. C. Ho, Yufan Zheng · Asian Control Conference · 2004
In this paper, a training and pruning algorithm is proposed for feedforward neural networks based on Jacobian rank deficiency. Redundant nodes are pruned during the training process, and computational cost of the training algorithm is reduced significantly. Simulations are presented to demonstrate the effectiveness of the proposed approach.