Newsgroups: comp.ai.neural-nets
Path: cantaloupe.srv.cs.cmu.edu!nntp.club.cc.cmu.edu!hudson.lm.com!netline-fddi.jpl.nasa.gov!elroy.jpl.nasa.gov!usc!crash!pzcc.bitenet!news
From: Duane DeSieno <duaned@cts.com>
Subject: Re: Question; Real world data on a BP net
Organization: /etc/organization
Date: Thu, 29 Dec 1994 05:42:03 GMT
Message-ID: <D1K563.M7M@crash.cts.com>
References:  <xK8UPjv.jpeyton@delphi.com>
Sender: news@crash.cts.com (news subsystem)
Nntp-Posting-Host: loci.cts.com
Lines: 19

> I'm really new to neural networks (and probably vastly undereducated)
> I've been toying with an algorithm I've written for simple back-prop
> nets and it seems that one can only solve problems reduced to
> vectors of binary elements. I would like to know about anyone who has
> used BP nets to solve problems involving wide ranges of data. Perhaps
> a pre-processing technique or somethin'. Looking to model stock/futures
> data (the Holy Grail).
>  
>                         Appreciate any help. . .

One fun aspect of coding a neural network is the debugging process.  
The network can appear to work when there is a seriois error in coding.
It sounds like you have such a problem.  There is a lot of Software
available that has been tested.  You may want to buy instead of build.  

Duane DeSieno
Logical Designs
(619)459-6236
duaned@cts.com
