Re: neural nets

From: Russell Wallace (
Date: Wed Jan 25 2006 - 06:45:15 MST

On 1/25/06, CyTG <> wrote:
> I'd even be concerned with the inherent limitations on real-number number
> crunshing in a 32/64bit processor(cost effective of course, you wanna do it
> natively, not carry alot of carry's in code).
> If you say thats not an issue, then fine :) .. could be cool if you could
> visualize *why* it aint an issue( maybe it is just an assumption so
> intuitively correct, it need no proof!)

64 bit floating point gets you about 16 decimal digits of precision, which
is much better than you get from analog/semi-analog devices like biological
neurons; AFAIK there is no device other than a digital computer that can do
better, which strongly suggests precision isn't a problem here.

- Russell

This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:55 MDT