Home › Neural Network Forums › Neural Networks › Getting Started ›

Some question on Encog Engine

sujan's picture
Submitted by sujan on Fri, 03/19/2010 - 18:41
in

what is the default learning rate for the back propagation learning in Encog? and
what is the gradient momentum?

do you guys used optimum values of learning rate and gradient momentum for Multilevel Perceptron.

please let me know...

‹ A question for a NN project OCR applet help ›
jeffheaton's picture
Submitted by jeffheaton on Sun, 03/21/2010 - 06:16.

There are no defaults for backprop in Encog. Both momentum and learning rate should be specified in the constructor to the backprop class. I usually start at around 0.5 for each and work my way from there.

You might also want to look at Resilient propagation (RPROP), it is generally much faster than backprop.

Copyright 2005 - 2012 by Heaton Research, Inc.. Heaton Research™ and Encog™ are trademarks of Heaton Research. Click here for copyright, license and trademark information.