Drexel dragonThe Math ForumDonate to the Math Forum



Search All of the Math Forum:

Views expressed in these public forums are not endorsed by Drexel University or The Math Forum.


Math Forum » Discussions » Software » comp.soft-sys.matlab

Topic: Neural Network performance normalization
Replies: 3   Last Post: Feb 8, 2013 9:40 AM

Advanced Search

Back to Topic List Back to Topic List Jump to Tree View Jump to Tree View   Messages: [ Previous | Next ]
Greg Heath

Posts: 5,944
Registered: 12/7/04
Re: Neural Network performance normalization
Posted: Feb 7, 2013 7:02 AM
  Click to see the message monospaced in plain text Plain Text   Click to reply to this topic Reply

"EanX" wrote in message <ket9j8$17c$1@newscl01ah.mathworks.com>...
> In designig a MLP NN I noticed that, for the inputs that I'm considering, I obtain a Performance value of 1.05e3 and this value become greater as inputs size increase, so I have to set
>
> net.performParam.normalization='percent';
>
> in order to have a performance value in range [0 1].
> Is this correct?


I don't know what you mean by performance value.

I recommend ALWAYS standardizing your inputs

[ xtrn, mu, std ] = zscore(Xtrn')';
xval = (Xval-mu)./std;
xtst = (Xval-mu)./std;

AND, standarizing your outputs if they are not binary.

If your current "performance value" is MSE , then normalize it with
MSE00 = mean(var(T,1)) to get

NMSE = MSE/MSE00
R2 = 1-MSE/MSE00

See the recent post on ANN_Error Goal

Hope tis helps.

Greg



Point your RSS reader here for a feed of the latest messages in this topic.

[Privacy Policy] [Terms of Use]

© Drexel University 1994-2014. All Rights Reserved.
The Math Forum is a research and educational enterprise of the Drexel University School of Education.