Biological Chip Coatings

Bill Armstrong arms at cs.UAlberta.CA
Fri May 17 22:57:38 EST 1991

rowland at ponder.csci.unt.edu (Keith Rowland) writes:

>In article <1991Apr2.164740.28858 at fcom.cc.utah.edu> strohsch at mines.utah.edu (David A Strohschein) writes:
>>  I am trying to find information about growing nerve cells on integrated
>>circuit surfaces.  The object is to grow the nerves on the ICs while 
>>retaining the normal or near normal electrical and physiological functions of
>>the nerve tissue.

>   While not actually growing neurons on ICs we do grow them on glass plates
>on which a grid of 64 electrodes has been photo-etched in an area of approx.
>Keith E. Rowland                rowland at cnns.unt.edu

In our work on adaptive logic networks (ALN) we have come a long way
in getting them to find simple rules in boolean data.  If you could
get sequences of bits representing axonal pulses in your collection of
neurons, you could try to predict what causes what by taking a many
samples of all signals BUT x over the past t seconds, and training an
ALN to predict the current x.  I suppose you have tried this with
backpropagation networks, but we have found that ALNs are very good at
finding simple rules for generalization.  For example ALNs have
learned (a very good approximation for) the rule for a multiplexer
with 521-input signals, based on 8000 samples, an infinitesimal part
of the whole space.  So it might be worth while to try this out on a
collection of neurons.

Your discipline, about which I know virtually nothing, probably has
very sophisticated ways of studying interaction of neurons, so maybe
the suggestion is not useful.  In any case, I'd appreciate your
comments on the idea.

Bill Armstrong

P.S.  The ALNs are on menaik.cs.ualberta.ca [] in
Prof. William W. Armstrong, Computing Science Dept.
University of Alberta; Edmonton, Alberta, Canada T6G 2H1
arms at cs.ualberta.ca Tel(403)492 2374 FAX 492 1071

More information about the Neur-sci mailing list

Send comments to us at biosci-help [At] net.bio.net