Since a number of Sci.Skeptic readers seem to be into GAs, this seems like a good cue to m
Organization: Knowledge Systems Lab, Stanford University
From: rice@sumex-aim.Stanford.EDU (James Rice)
Since a number of Sci.Skeptic readers seem to be into
GAs, this seems like a good cue to mention a great new book
on this subject:
Genetic Programming: On the Programming of Computers
by Means of Natural Selection.
by John R. Koza, MIT Press 1992.
This is actually being shipped by MIT press as of today.
I admit to having had something to do with its production,
so I am biased, but I am told that apart from this being
a pretty important book in its field, it also contains
about the best explanation of Holland's Slot Machine and
the Schema theorems, which are the basic mathematical
underpinnings which demonstrate why evolutionary genetics
actually works (and is actually near optimal in some senses).
I seem to recall a post from some dipshit only a couple of
days ago saying "~evolution doesn't work." Well, aside from
the obvious and well documented paleontological evidence
and bacterial examples from the real world, it turns out that
there's a whole bunch of theory and practice from the
computer science area that rather elegantly demonstrates
evolution. Maybe we need a bit more cross fertilisation
between comp. sci. and biology.
For those interested in the original theoretical work,
this is all covered in:
Adaptation in Nature and Natural and Artificial Systems
by John Holland. Univ of Mich Press 1975 {and I believe
a second edition 1992 from MIT press}
However, I gather that most people find Holland's text
pretty well unintelligible. I've never had the strength
to read it, having been warned off it. I'm content to
believe that his maths it better than mine.
For those who would just like to get a quick and easy
idea of what Genetic Programming can do, I highly
recommend (Plug alert) the following movie:
Genetic Programming: The Movie
By John R. Koza and James P. Rice, MIT Press 1992.
It's an hour long, snappy, intense and has good production
values (no, I don't get a royalty).
Rice.
E-Mail Fredric L. Rice / The Skeptic Tank
|