in backpropagation learning algorithm always i used to get the same accuracy rate whatever learning rate and momentum constant is used.So, what is the solution for this

whatever value for learning rate is given the BPN produce the same accuracy rate.what to do

Answer 1

[Greg Heath]

I have the same experience.

What I do is use the best of multiple designs that differ by number of hidden nodes and initial random weights.

For examples search both NEWSGROUP and ANSWERS using

 greg fitnet

for regression & curvefitting

and

 greg patternnet

for classification & patternrecognition

Hope this helps.

Thank you for formally accepting my answer

Greg