|
Authored by: JohnF on Saturday, May 11 2013 @ 12:16 AM EDT |
Well at least they do not appear to always operate in a
predictable manner like a computer. They do not appear to be
entirely deterministic! (Human brains that is!)
Regards
John F[ Reply to This | Parent | # ]
|
|
Authored by: JohnF on Saturday, May 11 2013 @ 12:24 AM EDT |
One further point. It depends upon how you define a
computer. So far the conventional definition is a box of
bits manufactured by humans and programmed
deterministically. The normal manner of testing a computer
is to use a set of standard data and feed it in together
with the written program and see what comes out. It should
be the same each time. Human brains are not quite like
that. Being chemically based cells that produce electrical
charges - things are never the same from one second to the
next, and it is not possible to repeat a calculation exactly
each time!
Regards
John F[ Reply to This | Parent | # ]
|
|
Authored by: jesse on Saturday, May 11 2013 @ 06:59 AM EDT |
Not really. The problem is the number of states.
In any simulated neural net the number of states is very definitely finite - the
number of bits in every value. VERY large, but still finite.
A real neural net (either from living tissue, or through a complex structure of
memristors, has an infinite number of states (the state of every atom/chemical
molecule..), and is influenced by quantum level effects.
Second, computing the final state of any living neural net (or that complex
structure of memristors) is only a statistical result - you can never get a
given neural net back to the same state it was before the measurements. Nor can
you exactly duplicate the state of one neural net into another. You can get
close...
Simulated neural nets? no problem. Just load the same bit values again.[ Reply to This | Parent | # ]
|
|
|
|
|