the value of spread must be varied from 2 to 0.1?

I Varried the value of spread in newrb, to get the best performance of training with radial basis function network.But, i found the same results for all different value of spread.it 's possible?

Answer 1

[MD QIDWAI]

If you are using your sample data/Training data for comparing performance, then you will not get any effect of spread value. However, divide your sample in two groups, for training and verification. After training of the RBN, use your other data for verification. You will get to see the effect of spread.