Re: [computer-go] Data Mining

["Frank de Groot" <frank@xxxxxxxxxxxxx>]

----- Original Message ----- 
From: "Mark Boon" <tesuji@xxxxxxxxxxxxxxxxx>
Subject: RE: [computer-go] Data Mining


> > My setup is very crude and the amount of data per game
> > collected is limited (# of games is large).  Does anyone have a
> > similar setup that they use for data gathering that might be more
> > useful?  I am hoping to avoid effort in enhancing my current setup
> > by simply running someone else's tools and returning a copy of
> > the output for there use as well.
>
> Maybe Frank can hook up with this guy? Sounds like a perfect combination
if
> you ask me.


When I first designed m pattern learning system, I thougth I needed 64
Terabyte (better still 256 TB).
Because to keep score of a matrix of 8 Mpatterns and their relative values
would require that.
(So that you can see which patterns are how often preferable to which
others, and how often they aren't).

But I found a method to do it in just a third of the RAM in my PC.

Similarly, I think 3 TB is too much for anything, it's just the algorithms
and data mining methods used that are not clever.

_______________________________________________
computer-go mailing list
computer-go@xxxxxxxxxxxxxxxxx
http://www.computer-go.org/mailman/listinfo/computer-go/