Newsgroups: comp.ai.neural-nets,comp.theory.dynamic-sys
Path: cantaloupe.srv.cs.cmu.edu!nntp.club.cc.cmu.edu!hudson.lm.com!news.pop.psu.edu!news.cac.psu.edu!howland.reston.ans.net!torn!watserv2.uwaterloo.ca!news
From: Chris Elias <cdelias@watarts.uwaterloo.ca>
Subject: ***Q:Connectionist Nets and Dynamic Systems Theory
Message-ID: <CyLwnn.MKu@watserv2.uwaterloo.ca>
Sender: news@watserv2.uwaterloo.ca
Nntp-Posting-Host: charlemagne.uwaterloo.ca
Organization: University of Waterloo
Date: Tue, 1 Nov 1994 20:52:35 GMT
Lines: 21
Xref: glinda.oz.cs.cmu.edu comp.ai.neural-nets:19815 comp.theory.dynamic-sys:1973


Clearly, neural/connectionist networks are dynamical systems by any
definition.  Furthermore, networks can exhibit all kinds of dynamic
behaviour (i.e. chaos, catastrophe, bifurcation, etc.)  I was 
wondering if anyone is aware of a proof or discussion of the 
equivalence of general dynamic systems and neural networks.

In other words, is it possible to 'translate' any dynamic description
of a given system to a neural network description  (i.e. from
a system of DEs to a system of nodes)??

thanks in advance,
chris elias

-----------
Computational Epistemology Lab
http://beowulf.uwaterloo.ca/
Chris Elias
cdelias@watarts.uwaterloo.ca


