Abstract
The backpropagation algorithm for training feedforward neural networks is extensively used to solve pattern recognition, signal processing and control problems. However, the time required before convergence is long even for medium sized network problems. The choice of learning rate η and momentum coefficient α also have significant effect on the rate of convergence. In this abstract, a fast training algorithm for feedforward neural networks is briefed. This algorithm is based on the adaptation of learning rate by correlation coefficient. The algorithm has been extensively tested. Results show that this algorithm can significantly reduce the time required for convergence.
Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
