Subject:
|
Re: an idea, can someone tell me if this is possible/been done before/etc?
|
Newsgroups:
|
lugnet.robotics
|
Date:
|
Wed, 3 Dec 2003 03:11:44 GMT
|
Original-From:
|
Gordon Elliott <gelliott@csiscSTOPSPAM.cc>
|
Viewed:
|
921 times
|
| |
|
|
Ah, I do suspect that my data was a little out of date. However the
information I see (http://www.ccwu.edu/Thesis_Moynihan/Chapter%205.doc
Microsoft Word ".doc" file, or
http://www.pc.rhbnc.ac.uk/zanker/teach/Ps306/L2/Ps306_2.htm ) list the
housefly at 10^5 or 100,000, to 10^6 or 1 million neurons. This is a little
bigger than "400 neurons." In fact the housefly eye has over 3000 segments,
so its vision system alone is much larger than that. Remember that each
neuron has many synapses, so simulation requires handling ALL the synapses
of each neuron and that multiplies to 10s and 100s of millions of operations
per time resolution of the synapse for 10^6 synapses.
Still a pretty fair collection of RCX processors, but possibly not
supercomputer scale today.
-- Gordon
----- Original Message -----
From: "Steve Baker" <sjbaker1@airmail.net>
To: "Gordon Elliott" <gelliott@csisc.cc>
Cc: <lego-robotics@crynwr.com>
Sent: Tuesday, December 02, 2003 7:00 PM
Subject: Re: an idea, can someone tell me if this is possible/been done
before/etc?
> Gordon Elliott wrote:
> > On "complexity", I would point out that the computational power of the
> > neural pathways of the house fly thoroughly exceeds the computational
> > throughput of our most massive supercomputer clusters at the moment.
>
> That was true 20 years ago.
>
> A housefly's brain contains just 400 neurons. Your PC can easily simulate
> a 400 neuron neural network at faster update rates than biological • neurons.
>
> The problem isn't the raw computational power.
>
> ---------------------------- Steve Baker -------------------------
|
|
Message is in Reply To:
4 Messages in This Thread:
   
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
