IUBio

Trying to develop a computer model of biological neural networks

david_olmsted at my-dejanews.com david_olmsted at my-dejanews.com
Thu Sep 10 21:13:50 EST 1998


Ray Storm said:

"I’m very much fascinated by biological Neural Networks. But because I’m
a computer scientist, and not a biological or medical, I don’t know a
great deal of the biological aspects of Neural Networks. I’ve made a very
simplified requirement specification of a computer model for Neural
Networks.
I would like to ask someone to take a look at it (form a biological point
of view), and tell me what else I need to know, or what’s wrong in it.

Any other comments are welcome…

Thanks, Ray"

Dear Ray,

So good to hear from one interested in developing biologically realistic
neural networks. I agree with Matt Jones that the standard neuron model
having the threshold, etc. which you describe and which we all learned in
school is most likely not realistic. The alternative as the information
processing locus is the microcircuit consisting of groupings of synapses
which was first championed by Gordon Shepherd 20 years ago. This leaves the
neural threshold as simply a neuron's way of transmitting analog values over
great distences via action potentials. I must also add that neural networks
based upon the standard neuron model have many limitations in practice such
that they are now called artificial neural networks (see the History of
Neural Networks at my site below).

Books giving the details of realistic neuron computer simulations have been
authored by Christof Koch and are available from Amazon.com. He is coming out
with a new book considering the very latest information in November.

If you want a slightly higher level approach to neural network simulation
consider my site at http://www.neurocomputing.org. Here I assume all the
non-linearities of neurons cancel out since they are all proportional to each
other in the same fashion that the non-linearities of transister physics
cancel out in circuit analysis. With this assumption I have proposed that
asynchronous multivalued logic is the theory behind the microcircuits and
show several different neural networks based upon that assumption meant to
represent different regions of the brain. While almost all theorists are
working from the small upwards I am working from the behavior downwards.

Sincerely,

David Olmsted







-----== Posted via Deja News, The Leader in Internet Discussion ==-----
http://www.dejanews.com/rg_mkgrp.xp   Create Your Own Free Member Forum



More information about the Neur-sci mailing list

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