In article <ANDRZEJ.93Jul6161824 at bcars182.bnr.ca>, andrzej at bnr.ca (Andrzej Bieszczad) writes:
|> >>>>> On Tue, 6 Jul 93 18:55:02 GMT, ip001b at uhura.cc.rochester.edu (Ivan Pulleyn) said:
|> IP> Nntp-Posting-Host: uhura.cc.rochester.edu
|>|> >(well... in recent years) people have realised that there are things neural
|> >nets can do better than the digital computer, and it's revived NN research.
|> IP> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|>|> IP> I am confused by what you are refering to here. What do NN's run on,
|> IP> besides digital computers? All parallel processing can actually be
|> IP> implemented on serial machines, so that can't be it.
|> IP>
|>|> IP> Ivan.......
|>|> NNs are SIMULATED on digital computers.
|>|> Ideally, NNs should be analog machines (... that takes us quite far
|> back to the past). Unfortunately, the technology is not yet in place
|> for that. I am a true believer that it is only a matter of time,
|> though.
|>
[more stuff cut]
I fink Alan Turing showed that provided you had a big enough digital computer,
enough time, etc, you could simulate just about anything on it.
You'll have to excuse me if that's wrong - long time, bad memory, etc.
Yup... most NN work is done by simulation on digital computer (I believe).
The point is that what takes a digital computer a night to simulate can be
done in milliseconds by a NN.
There are NN chips becoming available now - can't point you at em, cos this is
all way outside me area of expertise.
Oh... and analogue computers is still in use, although rarely encountered.
One use is is seizmic work, I believe.
Danny (still way out of any area he knows anything about - so no
flames please :)
--
Danny Banks. eep1db at ee.surrey.ac.uk
-------------------------------------------------------------------------------
I'm *not* a mad scientist............................I'm an eccentric engineer.