Newsgroups: comp.ai.neural-nets
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!howland.reston.ans.net!cs.utexas.edu!utnut!oci!owl.sunnybrook.utoronto.ca!farrell
From: farrell@owl.sunnybrook.utoronto.ca
Subject: Question: Should 1 or 2 outputs Make a Difference?
Message-ID: <3NOV94.19210355@owl.sunnybrook.utoronto.ca>
Sender: news@oci.utoronto.ca
Organization: MIT PLASMA FUSION CENTER
Date: Fri, 4 Nov 1994 00:21:03 GMT
Lines: 30

Hi Netters,

I am a graduate student and relatively new to this field.  I am using a
neural net on SPECT imaging data and training the net to distinguish
between subjects with or without a behavioural problem (namely, Hemispatial
Neglect).

My question is:

        If the network is learning to classify a binary variable, such as
the presence or absence of some behaviour, should it make a difference whether
the target output is a single output ranging from 0 to 1 or a dual output
where "01" would be one classification and "10" would be the other possible
category?

I've tried it both ways and have found that my network with 2 output units
performs worse than my network with only 1 output.

Should this be so, and if yes why?

I would appreciate any help anyone would care to offer, whether it be a
direct answer or a point to a good reference ;-)

Thanks to all in advance,

Farrell Stuart Leibovitch
Graduate Student at the Institute of Medical Science
E-mail farrell@owl.sunnybrook.utoronto.ca

Kindly e-mail me directly so as not to add traffic to this newsgroup
