Re: [sl4] The Jaguar Supercomputer.

From: Luke (wlgriffiths@gmail.com)
Date: Sun Nov 22 2009 - 14:12:43 MST


As I understand it, any given synapse can have an enormous range of
"weights", from negative through positive. You never have a "if this neuron
fires, then that neuron fires" situation, right?

 - Luke

On Sun, Nov 22, 2009 at 3:42 PM, John K Clark <johnkclark@fastmail.fm>wrote:

> On Sun, 22 Nov 2009 "Luke" <wlgriffiths@gmail.com> said:
>
>
> > Matt Mahoney: Where does this 10**9 (10**15) come from again?
> > Is that the full storage capacity of the brain?
> > Something like (num_synapses * num_possible_weights_per_synapse)?
>
> I'm not Matt Mahoney but I'll give it a try. Assuming as seems likely
> that memory works by Long Term Potentiation (LTP), you can make an
> estimate of the maximum storage capacity of the brain. There are about
> 10^11 neurons in the brain and each neuron has about 10^4 synapses. I
> have not seen any evidence that LTP can store more than one bit so that
> gives us a figure of about 10^15 bits or 10^14 bytes of storage capacity
> for the human brain.
>
> However this is almost certainly a considerable overestimate, I don't
> know the true figure but it must be less than that. In the January 28
> 1994 issue of Science Dan Madison and Erin Schuman found that LTP
> spreads out over a large area so you have lots of copies of the same
> thing. Neural Net expert Terrence Sejnowski said of this "Instead of
> thinking of a synapse as representing a piece of information you can now
> begin thinking of a population of potentiated synapses acting together".
>
> John K Clark
>
>
>
>
> in particular
> the work of In the January 28 1994 issue of Science.
> They
> --
> John K Clark
> johnkclark@fastmail.fm
>
> --
> http://www.fastmail.fm - A no graphics, no pop-ups email service
>
>



This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:01:05 MDT