RE: computer-go: Pattern matching

[birk@xxxxxxxxxxxxxxxxxx]

jmassung@xxxxxxxxxxxxxxxxx wrote:
> People learn by knowing the basic rules of the game, and when they first sit 
> down to play, they have some basic ideas as to what might win the game 
> (capture stones for example) and they know from the rules how to do it 
> (surround the group to continue the example) and they do have some basic 
> patterns they probably saw in the rule book.

> Everything else is learning, though.  And in my mind, a "true AI" Go program 
> would be able to edit its own pattern database.

That is not (entirely) true. Most 'pattern' a player known he did
not find by himself, but was taught by a stronger player.
E.g. I did not re-invent the bamboo-joint, it was shown to me as good
way to connect.

> 1. Most obviously - how is a pattern chosen? Typically, this is something 
> that happens when someone gets burned in a game.

I disagree. I think it's the other way around. Good patterns are 
learn from good (NOT bad) example ... or
from somebody who explained them.
How many players have deduced on their own the weakness of an empty triangle?

> 2. How does the program determine which pattern is bad after losing? 

that's why you (even a human) cannot deduce 'bad' patterns from a lost
game. You might be able to learn that a particular pattern did not
work as expected. Pattern just suggest moves (strategical/tactical 
'short cut') but don't relieve you from the necessity of reading ahead.

Christoph