From: Ben Goertzel (ben@goertzel.org)
Date: Mon Jun 04 2001 - 13:17:08 MDT
> Ben Goertzel wrote:
> >
> > Hi all,
> >
> > One of the several avenues I'm exploring in trying to find a way to
fund
> > Webmind AI research on an ongoing basis is *bioinformatics*
> >
> > I believe that the Webmind AI Engine can provide an unprecedentedly
> > powerful way of processing gene and protein expression data as
> produced by
>
> What makes you think built-in artifishl intelligence for pattern
> matching is noticeably better than monkey intelligence doing an
> interactive session with a frontend to serious (it depends on
> the problem set, usually a fat PC would do plenty, but of
> course a Beowulf can be useful, too) crunch?
a)
The extensive existing research literature on the topic
b)
A look at the data itself. It's big, it's nasty, it's messy.
Attached here is a document reviewing previous approaches to the
problem,
and explaining (in mathematical, not implementational terms) how Webmind
can
be used to surmount the obstacles that have been found. If you are
seriously interested in discussing this, please follow some of the
numerous
hyperlinks referenced there and read the papers at the end of them. The
actual data files are also linked to.
Within the WM team, we have already progressed a little beyond the
general
ideas in this document, but the further progress has been more technical
and
thus not appropriate for this genreal audience.
As for the "interactive" approach -- yes, it's clear that the best
approach
to this problem is a mix of AI with IA (computer based intelligence
augmentation). both the human mind and the AI mind can bring their own
unique power to the problem, is my intuition and also that of many
researchers in the field.
Ben
This archive was generated by hypermail 2.1.5 : Sat Nov 02 2002 - 08:07:57 MST