Newsgroups: comp.ai.neural-nets
Path: cantaloupe.srv.cs.cmu.edu!rochester!udel!news.sprintlink.net!redstone.interpath.net!sas!mozart.unx.sas.com!saswss
From: saswss@hotellng.unx.sas.com (Warren Sarle)
Subject: Re: Number of output neurons in classification
Originator: saswss@hotellng.unx.sas.com
Sender: news@unx.sas.com (Noter of Newsworthy Events)
Message-ID: <DBu44M.AH2@unx.sas.com>
Date: Mon, 17 Jul 1995 00:31:34 GMT
X-Nntp-Posting-Host: hotellng.unx.sas.com
References: <DBq49K.Hxq@murdoch.acc.Virginia.EDU> <DBruHo.IGp@murdoch.acc.Virginia.EDU>
Organization: SAS Institute Inc.
Lines: 20


In article <DBruHo.IGp@murdoch.acc.Virginia.EDU>, "Vijay S. Desai" <vsd3s@virginia.edu> writes:
|> ...
|> I forgot to mention in the earlier post that in case of multiple
|> categories using the "softmax" function proposed by Bridle does make
|> sense in which case one uses n neurons if there are n categories. My
|> question was does using the "softmax" function add much value if one
|> has only two categories, because the "softmax" function will require
|> the use of two neurons.

If you have two output categories, using softmax with two output units
is equivalent to using the usual logistic activation function with
only one output unit, since one of the two output units need not
be connected to anything else except a bias.

-- 

Warren S. Sarle       SAS Institute Inc.   The opinions expressed here
saswss@unx.sas.com    SAS Campus Drive     are mine and not necessarily
(919) 677-8000        Cary, NC 27513, USA  those of SAS Institute.
