[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re[2]: Intelligence



P.J.Leonard wrote:

> Robert Jasiek wrote:
> 
>> Divine intelligence
>>      is the ability to unify knowledge representation and application.
>> 
>> Example: Select the right function f. Then f(situation) = perfect_move.
>
> Well put  . .  .  now write function f()    :-)

Einstein said of all the work he had done, that he did it because he wanted to
read the mind of God.

Fortunately, we are permitted to start with an easier task; all we must do is
to read the mind of man.  Each pro has his _own_ function f() which, although
it does not always provide the perfect move, performs well in the long run.

So, even if we have to live with a few false positives (occasional bad moves)
it may be possible to duplicate some pro's function f(), and use it to generate
_some_ move, even if not the perfect one, and even, ultimately, to have the
program emulate the _style_ of one particular pro or another.

Of course, representing the pro's knowledge is a daunting task, in and of
itself.  (Not as hard as reading God's mind, I guess.)  This sort of thing has
already been done in other areas of discipline.  For example, there are
programs which can write sonatas in the _style_ of Beethoven, or in that of
Bach, those styles having been encoded as production rules in some grammar that
represents the styles' features.  This encoding is not done "by hand", but by
supplying as input many examples of the style, and then allowing the program to
observe and extract the features of the style, from this data set.

What is required is to uncover the _syntactic_ patterns of an individual's go
behavior.  See previous messages from me to this forum, for a bibliography of
texts which, I hope, will prove useful toward this end.  In particular, let me
repeat the following endorsement:

> A Probabilistic Theory of Pattern Recognition 
> Luc Devroye, Laszlo Gyorfi, and Gabor Lugosi
> Springer Verlag; ISBN: 0387946187
> 
> The table of contents alone is worth reading.  It's at Devroye's web site:
> 
> http://www-cgrl.cs.mcgill.ca/~luc/pattrec.html

This excellent text seems to hold the key to understanding, if not the mind of
God, then at least the mind of man.  The difficulty, as always, lies in how one
will represent the data.

Rich
-- 
Richard L. Brown              Office of Information Services
rbrown@xxxxxxxxxxxxxxxxx        University of Wisconsin System Administration
rlbrown6@xxxxxxxxxxxxxxxxx    780 Regent St., Rm. 246  /  Madison, WI  53715