Re: computer-go: speculative introspection

[Heikki Levanto <heikki@xxxxxx>]

William Harold Newman <william.newman@xxxxxxxxxxxxxxxxx> wrote:
> How it is that I can step up to a Go board and recognize quickly that
> there's a solid white connection between point A and point B? Even
> when the connection involves tens of stones, my wetware seems to
> notice it in perhaps 500 milliseconds. 


You could have two images of the board, one with the actual stones, and one
with a "simplified" picture of it. Between these two you'd have a huge
number of pattern matchers, that convert a connecting pattern into a solid
connection. Then all you have to do is to trace the path of solid stones
from A to B.  There may be multiple levels of this, first recognising likely
cutting patterns and their failures, and only then simplifying.

. . . . . .
A . X . . X
. . X . . .
. . . . X .
. . . . . X
. . . . . .
. . . . . .
. . . . B .


X . X   -> X X X
. . X      . . X

. . . .      . . . .     . X X .     . . . .
X . . .  ->  X . . X  -> X . . X ->  X X X X
X . . X      X X o X     X X . X     X X X X
. . X .      . o X .     . o X .     . . X X

X .         . X . .      X .
. X    ->   . . X .  ->  X X
. .         . X O O      X .
. .         X . X .      X .
X .         . X . .      X


. . . . . .
A x X . . X
. . X x x .
. . . . X x
. . . . x X
. . . . x .
. . . . x .
. . . . B .


All the possible patterns can be matched in parallel, so they should happen
within a few "clock ticks". 

I do not actually propose that this is the mechanism used by your brain, but
it could be something like this.  One way to study it would be to arrange a
set of test positions, with easier and mroe difficult connection, some that
are connected, and some that are not. Measure the time it takes to classify
them, and error rates. Plot against go experience of the test subjects.

I guess this is a close analogy to how we sort our visual input into a set
of objects. This must have evolved early on, and be highly optimized in the
brain. The real life problem is much harder, since we are dealing with
irregular shapes in three dimensions, and they can move around.

-H



-- 
Heikki Levanto     LSD Levanto Software Development   heikki@xxxxxxxxxxxxxxxxx
               "In Murphy we Turst"