Re: Q: How to apply a trained network and save the results?

(Që…ð@ÄØð@ÄÔ)
Tue, 2 Sep 1997 08:22:20 GMT+100 

> Date:          Mon, 1 Sep 1997 18:31:27 +0200 (MET DST)
> Reply-to:      snns@informatik.uni-stuttgart.de
> From:          "Juan P. Rigol" <jrigol@goliat.ugr.es>
> To:            kh@ier.uni-stuttgart.de
> Subject:       Q: How to apply a trained network and save the results?

Hello Juan,

> Hello all. I'm a new user of SNNS for Solaris and I
> was wondering if anyone out there could help me
> (I don't want to bother SNNS staff any more :)
> My question is how to apply my trained network to
> a new pattern file and save the output to a file.
> I've created three pattern files:
> 1 for training, 1 for validating, 1 with unknown output.
> I've defined my bp/feed-forward network topology (7:14.1)
> and trained it stopping when reaching the lowest error
> in the validating pattern file. At this point, I would like
> to classify the third pattern file with the trained network.
> I've tried to save a .res file but it seems it only works
> for the pattern file used for training. I suppose that if I
> select my third pattern file as the current pattern it will
> update my network activations when pushing ALL in the control
> panel. What should I have to do?

Just select the third file as training file at the control panel and 
then do not push the "ALL" button for training but just save a result 
file. You've got what you want the in that result file.

Regards,

Kai
!!! Bitte beachten: Ich/Wir haben eine neue e-mail Adresse:
!!! Please note: I/We have got a new e-mail address: (see above)
     ---> kh@ier.uni-stuttgart.de <---
It's a lot more convenient know !!!
Es ist wurde deutlich einfacher !!!

---------------------------<< End of Message >>------------------------------

Dipl.-Ing. K. Hufendiek, M.Sc.(UMIST)
Institut fuer Energiewirtschaft und
Rationelle Energieanwendung (IER)
Universitaet Stuttgart
70550 Stuttgart
Germany
Phone: +49-711-78061-58
Fax:   +49-711-7803953

http://www.rus.uni-stuttgart.de/UNIuser/ier/

-----------------------------------------------------------------------------