Path: utzoo!censor!geac!torsqnt!news-server.csri.toronto.edu!cs.utexas.edu!usc!zaphod.mps.ohio-state.edu!rpi!uwm.edu!bionet!agate!riacs!danforth
From: danforth@riacs.edu (Douglas G. Danforth)
Newsgroups: comp.ai.neural-nets
Subject: Re: Connectionist Finite State Machines -- description of an architecture
Message-ID: <1990Dec5.164602.8177@riacs.edu>
Date: 5 Dec 90 16:46:02 GMT
References: <7982@uwm.edu>
Sender: news@riacs.edu (James A. Woods)
Organization: RIACS, NASA Ames Research Center
Lines: 18

In <7982@uwm.edu> markh@csd4.csd.uwm.edu (Mark William Hopkins) writes:

>   This is a description of a rather simple architecture that can be used to
>train a neural net to be a finite state machine using only backpropagation.
		 (... rest of nice description deleted ...)

Mark,
    Your description brought back an interesting psychology controversy 
from the 60's.  Patrick Suppes at Stanford was able to show theoretically
that a Stimulus-Response model could be trained so that asymptotically it
would emulate a finite state machine.  Your architecture is now an embodiment
of that principle.  Thanks for the posting.

--
Douglas G. Danforth   		    (danforth@riacs.edu)
Research Institute for Advanced Computer Science (RIACS)
M/S 230-5, NASA Ames Research Center
Moffett Field, CA 94035