From: Ben Goertzel (ben@goertzel.org)
Date: Sat May 11 2002 - 10:59:53 MDT
> > Has
> > Novamente learned heuristics that enable it to connect locally visible
> > properties of a variable to its probability of distant entanglement?
>
> No, it hasn't learned any of its own heuristics at this point
On second thought, this sort of depends on one's interpretation of the words
involved...
It does adapt many parameters of given heuristics, to a given problem area
and to a given problem, which in a sense is "learning heuristics" from
within a restricted space of possibilities.
This is weaker than what I'd consider the default interpretation of the
phrase "learning heuristics"...
But on the other hand, even an app like EURISKO that is more transparently
"learning heuristics," is still doing so from within a fairly narrow range
of possibilities (defined by its initial operator set, in that case)
Learning heuristics from within a very broad range of possibilities has not
yet been demonstrated by any AI program ever, I'd say.
Of course, in theory the range is always bounded for any fixed memory size &
time span, though it becomes unbounded if one assumes one's AI has the
capability for physical self-modification and if one assumes appropriately
optimistic things about the physical universe ;>
ben
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:38 MDT