At 11:22 AM 3/28/2001 +0200, Eugene Leitl wrote:
> > don't think everyone is using it the same way.) Writing heuristics to
> > replace sections of the table is a sub-optimal solution (provably so) and
> > would have serious consequences if implemented on a broad scale.
>Actually, you can prove that that table is anything but random, unless
Actually, you can prove that an optimally compressed table will give the
most correct answers for all practical purposes. On average, heuristics
produce inferior results (on a couple different axes).
I am currently working on a paper that I hope to have completed and on the
web prior to Extro-5 that addresses many of the issues being talked about
here and proposes several new concepts that I think the people on this list
will find fascinating. No hand-waving, just fairly rigorous mathematical
formulations with lots of references. In particular, it addresses the fact
virtually every AI implementation attempted so far is provably defective in
one or more respects; apparently many AI researchers don't go dumpster
diving through mathematics archives like I do. I have intentionally
avoided really getting into this topic in email (even though I know a lot
more than I am stating) because the topic is way too long to do justice to
here and would necessarily be less rigorous than it should be.
This archive was generated by hypermail 2b30 : Mon May 28 2001 - 09:59:43 MDT