Newsgroups: comp.ai.neural-nets
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!fas-news.harvard.edu!newspump.wustl.edu!newsreader.wustl.edu!news.starnet.net!wupost!howland.reston.ans.net!cs.utexas.edu!utnut!cannon.ecf!srodula
From: srodula@ecf.toronto.edu (SRODULA  EWA ANNA)
Subject: [Q]: NN Recipes in C++ (Timothy Masters) & QUIT ERROR
X-Nntp-Posting-Host: spark23.ecf
Message-ID: <D43rF4.AzI@ecf.toronto.edu>
Summary: Relationship between QUIT ERROR and Maximum error?
Sender: news@ecf.toronto.edu (News Administrator)
Organization: University of Toronto, Engineering Computing Facility
Date: Thu, 16 Feb 1995 17:04:16 GMT
Lines: 17

Hello everyone,

I have been using (a recompiled 32-bit version) of NEURAL program
that accompanies the book "Practical Neural Network Recipes in C++"
by Timothy Masters.
Could someone explain the relationship between the QUIT ERROR parameter
(specified in the training script or command line) and the
maximum absolute error?
Context of my problem: I'm attempting to use a single hidden layer
network with 11 inputs, 5 hidden units and one output unit (ie.
the GENERAL model in the NEURAL program) to predict a time sequence
(an estimate of noise in a signal) and hence the concept of Maximum
Absolute Error is more relevant to me than MSE.

Any help would be much appreciated.

-Thas
