Share Email Print

Proceedings Paper

Near-optimal dynamic learning rate for training backpropagation neural networks
Author(s): Serge Roy
Format Member Price Non-Member Price
PDF $17.00 $21.00

Paper Abstract

Back-propagation neural networks is a very popular training algorithm for neural nets. One of the problems with this learning algorithm is its training speed. The selection of a good learning rate is a very important factor to achieve a satisfactory learning time. However, it is very difficult to determine an optimal learning rate since this parameter is dependent on a lot of variables such as the size of the network, the number of examples in the training sets... A new method is proposed to compute a near optimal learning rate for a three layer (one hidden layer) back propagation network.

Paper Details

Date Published: 19 August 1993
PDF: 7 pages
Proc. SPIE 1966, Science of Artificial Neural Networks II, (19 August 1993); doi: 10.1117/12.152627
Show Author Affiliations
Serge Roy, Defence Research Establishment Valcartier (Canada)

Published in SPIE Proceedings Vol. 1966:
Science of Artificial Neural Networks II
Dennis W. Ruck, Editor(s)

© SPIE. Terms of Use
Back to Top
Sign in to read the full article
Create a free SPIE account to get access to
premium articles and original research
Forgot your username?