Path: utzoo!attcan!uunet!lll-winken!xanth!nic.MR.NET!hal!cwjcc!tut.cis.ohio-state.edu!mailrus!shogun!msiskin
From: msiskin@shogun.cc.umich.edu (Marc Siskin)
Newsgroups: comp.sys.amiga
Subject: Re: AI with Ami...Neural Nets
Summary: Magellan has it somewhat
Keywords: neural, AI, gratuitous spam, Neural Nets
Message-ID: <963@mailrus.cc.umich.edu>
Date: 1 Mar 89 19:01:21 GMT
References: <9472@louie.udel.EDU> <5244@hcr.UUCP> <566@wsu-cs.uucp>
Sender: usenet@mailrus.cc.umich.edu
Reply-To: msiskin@shogun.cc.umich.edu (Marc Siskin)
Distribution: na
Organization: University of Michigan Language Lab, Ann Arbor
Lines: 22
UUCP-Path: mailrus!shogun!msiskin

In article <566@wsu-cs.uucp> jal@cs.wayne.edu (Jason Leigh) writes:
>
>You do not necessarily have to have a machine capapble of forking
>processes to build Neural Networks, although in some cases the result

>It depends on the kind of neural network you are trying to implement.
>Using message ports in neural nets seems a considerable overhead if
>not a waste since the only data that needs to be passed would be no
>more than a few bytes.
>
>I have build neural networks that solve travelling salesman problems
>that can be written in BASIC on a pocket computer...

According to the publisher of Magellan, it uses some Neural Networking
style programming.  Since my knowledge of NN is solely based on one
rather disjointed talk a year ago at a conference on Language Learning,
I can't vouch for how much Magellan supports Neural Net programming.

Disclaimer: I am not an employee of Emerald Intellegence or even a
satisfied customer of EI.  (I do have Magellan though, just haven't
had enough time to make up my mind about it).
         Msiskin@shogun.cc.umich.edu