Data Normalization for Artificial Neural Network use

Hello!! I would like to train my neural network with two customized equations in my hidden layers one similar to logsig and the other similar to tansig. My question is how to normalize my input and target data? Between [-1 1] OR [0 1] ?

Thank a lot for your feedback!

Answer 1

[Greg Heath]

For regression/curve-fitting just use FITNET. Defaults will take care of those details.

Typically, you only have to worry about an effective number of hidden nodes and an effective combination of random initial weights.

Check BOTH the NEWSGROUP and ANSWERS for some of my many posts.

 greg fitnet

Hope this helps.

Thank you for formally accepting my answer

Greg