Newsgroups: comp.ai.neural-nets
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!howland.reston.ans.net!agate!darkstar.UCSC.EDU!nic.scruz.net!earth.armory.com!rstevew
From: rstevew@armory.com (Richard Steven Walz)
Subject: Re: Scientific American
Organization: The Armory
Date: Mon, 26 Dec 1994 15:04:24 GMT
Message-ID: <D1FB7E.8M4@armory.com>
References: <3de8b7$3nr@gateway.ecn.com>
Sender: news@armory.com (Usenet News)
Nntp-Posting-Host: deepthought.armory.com
Lines: 22

In article <3de8b7$3nr@gateway.ecn.com>,
Tim Hess <tmhess@gateway.ecn.com> wrote:
>Scientific American Oct. 94 has an article in "The Amature Scientist"
>section called Building an Electron Neuron. (pg 136) They describe how to
>build a 'sun tracker' using photo cells and a 741 op-amp. In another article
>(Jan. '95) in letters to the editor, a writer argues it has been used since
>WWII. That the design is just a closed loop feedback control system. The
>original author (John Iovine) says you could hook "a hundred or so" op-amps
>together and teach it to play tic-tac-toe." Is it feedback, or is it NN?
----------------------------------------------
There was once an old math project using match boxes to store the token
array and you could teach it how to play perfect tic-tac-toe. There was
also a sixties or early seventies book called "We built our own
computers!", I believe British, which had a number of similar learning
engines in it. Teaching, i.e. closing the loop, is old in some types of
machinery. And yes, I re-invented the wheel for a heliostat with a couple
op-amps and a couple CdS photoresistors, and it worked fine. I even had a
relay circuit on it to reverse it and track it back to hunt the sun at high
speed if it came unlocked. Very simple. I used a hanging weight on a reel
to power the assembly and a solenoid brake.
-Steve Walz   rstevew@armory.com

