X-Message-Number: 26856
From: 
Date: Sat, 20 Aug 2005 15:04:15 EDT
Subject: Re: To Thomas Donaldson

You said:

1. We  cannot imitate a parallel machine with any smaller assemblage
of processors which may try to imitate subsets of the  processors
of the parallel machine. The original processors in  these subsets
can genuinely act at the same time, not at  different times no
matter how small. Fundamentally, a parallel  machine with N
processors cannot be accurately simulated by  any other machine
with M < N processors.
I disagree completly. A single processor can work out the same problems as  a 
multiprocessor system. A multiprocessor is simply faster for the same  
technological level.
I am surprised that a mathematician tell what you say, if you have any  
science reference on that I would be interested.


2.  Our neurons grow new connections with one another as part of 
learning. Sometimes our brain even produces new neurons for  the
same purpose. Various people on Cryonet (and elsewhere)  have
proposed that new connections might be formed instead by  light
beams between "neurons" or other nonbiological means. If  nothing
else, such a method will involve a lot of complexity  to actually
implement --- at least as much as we have with  brains. Growth
of new connections looks much simpler if we  have millions of
neurons.
Think about the Internet: it "grow" new connections continuously. If you  

think of each neuron as a microprocessor of a special kind, you have the  
answer.


3.  Before we think about making our brains faster, it would be  useful
to think just what use we would make of such speed.  
The aim is to have fast chips to process many neurons, one after the other,  
not to have a faster brain, at least for now.


Just  some comments on the project of building a "human brain".
I think the present objective is not to build a human brain, the first step  
would be to have a neuromorphic chip able to simulate any biological neuron or 
a  small network. The next step would be an application such a speaking 

translator  in real time, then would come a small mammal brain uploaded on a set
of 
chips,  the human project would be the next step.
 
I hope to have a broad map of the neuromorphic chip at the end of the year,  
a first sample may be two years away.

 
Yvan Bozzonetti.


 Content-Type: text/html; charset="US-ASCII"

[ AUTOMATICALLY SKIPPING HTML ENCODING! ] 

Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=26856