Newsgroups: comp.ai.neural-nets
Path: cantaloupe.srv.cs.cmu.edu!rochester!udel!gatech!news.mathworks.com!news.kei.com!simtel!news.sprintlink.net!redstone.interpath.net!sas!mozart.unx.sas.com!saswss
From: saswss@hotellng.unx.sas.com (Warren Sarle)
Subject: Re: Scaling the rows or the columns? With which method?
Originator: saswss@hotellng.unx.sas.com
Sender: news@unx.sas.com (Noter of Newsworthy Events)
Message-ID: <DFHJ30.Mnv@unx.sas.com>
Date: Mon, 25 Sep 1995 23:40:12 GMT
X-Nntp-Posting-Host: hotellng.unx.sas.com
References: <436auo$ekd@nx2.hrz.uni-dortmund.de> <437l6n$jhg@uuneo.neosoft.com> <Pine.SGI.3.91.950923091805.5374D-100000@bgumail.bgu.ac.il>
Organization: SAS Institute Inc.
Lines: 18


In article <Pine.SGI.3.91.950923091805.5374D-100000@bgumail.bgu.ac.il>, Zvi Boger <zboger@bgumail.bgu.ac.il> writes:
|>
|>      The standardization method should be preferred, as the outliers
|> are separated to the edges of the sigmoid, and does not influence much the
|> resolution of the rest of the examples (as would in min-max values
|> preprocessing). Also, the 0-1 normalisation uses only half of the sigmoid
|> range.

Except that in most types of networks, the inputs are multiplied by
adjustable weights and often combined with other inputs, so the
reasoning above does not apply.

-- 

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.
