[computer-go] Monte Carlo result as a feature for pattern matching

[Andrés Domínguez]

2006/7/8, Peter Drake <drake at lclark.edu>:
>
> We would like to have access to life/death and influence information
> in patterns.  For example, we would like to have patterns that say
> things like "if a weak friendly stone shares a liberty with a strong
> friendly stone, consider connecting them"

What "a weak friendly stone" means? Are you talking about a single stone?
Usually you shouldn't connect a single stone to a strong group,
probably you should make bamboo joint or other good shape, or leave
that stone.

> or "if a point adjacent to
> one of my stones is strongly influenced by me, playing on that point
> is probably not important".

I think that many programms do that. But I don't understand how to put
influence in patterns.

> The problem is that open-ended life checkers and influence detectors
> are quite difficult to write.
>
> I propose (and am working on) using Monte Carlo analysis to find the
> expected ownership of each point on the board (say, as a number in
> the range [-1.0, 1.0]) and offering this information as a feature in
> the pattern detector.

What do you mean with "ownership"? Probability of being one point at
the end of the game? Or balance of power on that point? The second one
looks like an influence function. If you mean "territory" I think that
it is not a good idea, since in the fuseki and middle game it's not
important each point of territory, but a global balance of territory.
I think that if you add the value for every point (0.4+(-0.2)+...) you
will get a big different result that real territory balance.

> Thoughts?

That's my humble opinion.

Andrés Domínguez