[from Gary Cziko 930116.0325 GMT]
Bill Powers (930114.1640) (>) responding to
Gary Cziko (930114.2200) (>>)
Genetic programming works by randomly creating (say 500)
computer programs using the functions and terminals believed
appropriate to the problem. The programs that come closest to
the criterion of fit are allowed to "mate" and their offspring
are similarly evaluated, etc. until a perfect or close
enough fit is achieved.
So there is something other than surviving to the age of
reproduction that determines whether mating takes place -- or
rather, something other than the final criterion of fitness
determines that survival.
You could make the criterion whatever you want it to be--it could be how
close the programs comes to typing out "PCT Forever" or coming up with a
program to predict closing prices on the stock market based on certain
economic indicators. It could also be how well a neural net (expressed as
a LISP program) is able to categorize and properly generalize perceptual
inputs. It is NOT the case that "something other than the final criterion
of fitness determines that survival." This remains the same throughout the
breeding process (although Koza HAS changed the criterion half-way through
a run to see how well the program evolution would be able to get back on
track--it does quite well).
This means that an external criterion
of fitness is being used: the experimenter's brain. It is the
judgement by the external party as to whether a given change
constitutes an "improvement" that determines survival, not
natural selection by failure or success at the task.
No, the criterion is how well the programs succeeds at the task (but, of
course, the task it what the experimenter's brain is interested in).
If now the experimenter could write into the program a way of
perceiving distance from the desired outcome...
This is what is done
and making the rate
of mutation depend on whether than distance was increasing or
decreasing, the criteria would be internal -- and you'd have my
model of the reorganizing system.
Sorry, mutation is not used in genetic programming (and not much used in
genetic algorithms either). "Sexual" recombination seems to provide all
the variation that is needed. And it is independent of how close the
population (or any given parent program) is to the goal. The variation of
the population remains quite constant throughout the run, although average
fitness does usually improve.
I think that part of the problem we're having with this "external" vs.
"internal" stuff is that Koza (and standard evolutionary theory) works on
populations edited by Darwin's hammer. The hammer is external, n'est-ce
pas? In contrast, you want the organisms to be able to "know" how well
they are doing so that they can "decide" whether to pass their genes down
intact (things are just fine) or scramble them up a bit (things are not so
good). This is not how Koza does it, and it seems to work out quite well
anyway.
This has got me wondering whether genetic programming could be used as a
model for reorganization within PCT. Lots of bad programs are tried out in
genetic programming before better ones are found and bred to evolve better
ones. If an organism did it this way, it wouldn't be around very long.
Maybe this is what you are trying to say to me. The reorganization process
itself must be "viable."
ยทยทยท
==============================================================
Greg Williams (930113)
Dr. Koza APPLIED for patents; it will be interesting to see if they are
granted. If they are, I suppose we will still be able to use his ideas to make
code for both educational and money-making purposes (which, of course, needn't
be mutually exclusive), presumably AFTER signing a license agreement with
Dr. Koza.
Greg, try this:
Koza, John R. 1990d. Non-linear genetic algorithms for solving problems.
U.S. Patent 4,935,877. Issued June 19,1990.
But we academic types probably don't have to worry too much about patents
like this since we don't usually make any money from our publications and
ideas. (Why would Koza sue us for patent infringement to get his cut if
the base was zero?) It's guys like you and Bill P. who try to make big
bucks selling programs that Koza will be on the lookout for.--Gary
P.S. Greg, can you tell me something about he book club from which you
obtained Koza's book? Do you get books cheaper than direct from the
publisher? Does it provide books that I would be interested in (I think
you have some idea of what my interests are by now outside of
PCT--evolution and selection theory as applied to any and all areas of
knowledge)?
------------------------------------------------------------------
Gary Cziko Telephone: 217-333-8527
Educational Psychology FAX: 217-244-7620
University of Illinois E-mail: g-cziko@uiuc.edu
1310 S. Sixth Street Radio: N9MJZ
210 Education Building
Champaign, Illinois 61820-6990
USA
------------------------------------------------------------------