From: Ben Goertzel (ben@goertzel.org)
Date: Fri Sep 09 2005 - 04:36:41 MDT
Mikko wrote:
> > An AGI needs a concept/hypothesis formation approach that plays
> well with
> > probabilistic inference, but I have never seen an adequate
> > concept/hypothesis formation method that's based solely on probabilistic
> > concepts. Do you have one that you'd care to discuss?
>
> Have you looked at independent component analysis?
Yes, because Ari Heljakka pointed me toward it. This technique seems to be
popular in Finland!
However, it is not oriented toward forming complex predicates satisfying
complex fitness functions (goals); I think its utility in an AI context is
probably restricted to perception processing, and general high-level
organization of concept-space...
What we use in Novamente for concept formation is a variant of BOA, which is
a probabilistically-defined variant of Genetic Algoriths / Genetic
Programming.
> It's not something I really know, but am acquainted with because my
> brother does research on that field. It would seem to be capable of
> hypothesis formation out of 'pure' data. I could be mistaken, though.
It is thus capable, but only of formation of restricted sorts of hypotheses.
-- Ben
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:52 MDT