X-Message-Number: 25163
Date: Sat, 27 Nov 2004 14:12:22 +0100
From: Henri Kluytmans <>
Subject: Simulated neural nets in FPGA's

Thomas Donaldson wrote :

>Even if FPGAs can be arbitrarily reprogramed, they still don't
>make it. You see, our brain is reprogramming itself, which 
>includes such things as neurons growing new connections and
>eliminating others. Furthermore, I doubt strongly that an 
>FPGA can be said to contain any "figurative" (ie. parts that
>work like) neurons at all. On this issue we are arguing matters
>of fact; if you simply gave me some references (EXPLICIT ones,
>as explicit as the ones I've just given you above), you might

Here is a reference to the work of Hugo de Garis, heading the Brain 
Builder Group at the Computer Science Department of Utah State 
University. I gave you a link more than one year ago :
(It seems you probably didn't bother to check it out.)

http://www.cs.usu.edu/~degaris/

The work concentrates on evolvable neural net models in 
programmable/evolvable hardware (FPGA's) to be scaled 
up to about a billion artificial neurons.

Please check out the publications on his website about his 
CAM-brains. And maybe then you will think different about 
being able to simulate (large enough) neural nets in FPGA's.
And, there you can read yourself, also the growing of new 
connections is simulated.

Grtz

Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=25163