No change in test performance by changing Learning Rate

hello all

I want to see the effect of changing Learning rates and Momentum on test performance of the neural network. But,even when I change the Learning rate ,test performance does not change.I am using 'trainrp' and 'trainscg' as training functions.(My aim is Pattern Recognition)

How is it possible?

Please help!!

Answer 1

[Greg Heath]

Why should the answer be different? Changing the learning rate should just change the number of epochs it takes to converge.

For what range of learning rates do you get the same answer?

What is the range of your inputs? How are they normalized in the net?

[piyush dugar]

Thanks for the help.

Range of learning Rate is from 0 to 1.

I've normalized using 'mapminmax' and 'removeconstantrows' .