Variable projections neural network training

8 The training of some types of neural networks leads to separable non-linear least squares problems. These problems may be9 ill-conditioned and require special techniques. A robust algorithm based on the Variable Projections method of Golub and Pereyra10 is designed for a class of feed-forward neural networks and tested on benchmark examples and real data.11

• Type: Journal
• Year: 2006
• Where: MCS
• Authors: V. Pereyra, G. Scherer, F. Wong
• Keywords: Feed-forward Neural Networks, MCS 2006, Neural Networks, Pattern Recognition, Variable Projections Method