[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [computer-go] Pattern matching - example play
At 11:58 PM 12/2/04, you wrote:
----- Original Message -----
Subject: Re: [computer-go] Pattern matching - example play
...
> this seems impressive. are you playing these moves in the same order?
Sorry, I do not understand the question.
I can take any pro game that the system has never seen before and on average
it predicts 46% of the moves, doesn't mattern in which order the positions
are presented.
ok, got it.
> i do play go (but only 1 dan). it's hard to imagine how you are going to
> "learn" timing and asking moves.
Because timing is irrelevant. It should simply play the best move.
yes, it should just play the best move.. the timing has to do with how
humans think about playing. the better players tend to leave ocal
situations unsettled if there are choices of ways to "settle it" and it's
not clear yet which way(s) might be more advantagous.
I don't know what "asking moves" are. (I don't play Go.)
a fairly urgent move that is made that forces a decision on the opponent.
e.g. a pincer will sometimes force an opponent to decide then whether or
not to invade a corner. or you will see an attachment made to one of the
stones that form a knight's move corner enclosure. the answering move can
basically say, i want the corner, i want the outside or let's fight about it.
> >The method of extracting the info is a trick of the trade...
> >that having no knowledge of Go oneself makes one doomed to failure..
>
> no it does not. it just seems that way to some of us.
Just refrain from telling me things like I am doomed to failure if I don't
become a Go player first, or that a pattern system won't play strong.
i'm obviously jumping into the middle of a long thread here, so i apologize
if i stepped on your toes. that was not my intention.
your prediction rate sounds great. does it vary much between the beginning,
middle and end stages of the game? does it seem to work better in differenr
styles of games like say between a small group to small group style versus
a sen-ren-sei or cosmic style (where most of the stones start out on the
4'th line in the begginning)?
...as I have
never intended such pattern system for anything more than Fuseki & Joseki.
The fact that it predicts a lot of pro moves is nice but it's a side effect.
The pattern system is also very useful to find patterns and look them up in
a database etc.
i see, that clears up some of my confusion.
It does not play Go well, and I am not surprized that it doesn't. A Go
playing program needs more than a sense of "good shape".
Nevertheless there is a continuous trickle of people emailing me "give up
for now - learn Go first" and "let me tell you, a pattern system will never
play Go". These kind of people go into my killfilter.
sure. but i suspect that most of them were trying to be helpful.
good luck with your research.
thanks
---
ray tayek <http://tayek.com/>, co-chair <http://www.ocjug.org/>, actively
seeking telecommuting work.
_______________________________________________
computer-go mailing list
computer-go@xxxxxxxxxxxxxxxxx
http://www.computer-go.org/mailman/listinfo/computer-go/