Newsgroups: comp.ai.neural-nets
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!howland.reston.ans.net!Germany.EU.net!nntp.gmd.de!news.rwth-aachen.de!newsserver.rrzn.uni-hannover.de!gm.pc!lzh
From: lzh@ifwmv2.ifw.uni-hannover.de (Christoph Graumann)
Subject: HELP - Back-Propagation Problem
Message-ID: <lzh.19.0010DA54@ifwmv2.ifw.uni-hannover.de>
Keywords: Neurograph, Simulation Laser Welding
Lines: 22
Sender: news@newsserver.rrzn.uni-hannover.de (News Service)
Organization: IFW
X-Newsreader: Trumpet for Windows [Version 1.0 Rev A]
Date: Fri, 30 Dec 1994 15:51:03 GMT

Hi everybody,

I try to design a neural network to determine the welding depth for a laser 
welding process based on experimental data. I am using a back propagation 
model with a binary input and about 20 neurons on the input layer. The hidden 
layer varies from 3 to 20 neurons. 

The training is quite perfect, but there are some problems left with new 
untrained parameter sets for which the NN has to predict the resulting welding 
depth of the process. I am interested in information and suggestions regarding 
the improvement of the network especially refering to the amount and size of 
hidden layers and the choice of training parameters. I am using the commercial 
available software NEUROGRAPH distributed by the university of Erlangen, 
Germany.

In case you can't help maybe you could let me know some further adresses for 
an inquiry. Your help will be greatly appreciated

Thanks sincerely    Christoph Graumann

e-mail :  lzh@ifwmv2.ifw.uni-hannover.de
fax      :  + 49 511 2788 100
