Talk:Vexbot
From PokerAI
The newest development by the University of Alberta games group is Vexbot, which is an AI system based completely on opponent modeling. This is a radical departure and very different from the original rule-based Loki. It has been successful thus far in proving that opponent modeling is critical when playing poker at a world class level. Vexbot forces an opponent to continually change strategies and adapt their play, as it will attempt to exploit any and all weakness or predictability it finds in their playing style. Players facing Vexbot find themselves asking "how can I win against myself?"
( quoted verbatim from http://www.poker-academy.com/media/poker-ai )
indiana: Hi, I see problem with the above text, it's written quite some time ago. So the "newest" development is no longer correct. Otherwise AFAIK Vexbot was derived from Poki (as said above, and Poki is successor of Loki), while Spartbot is derived from PsOpti. I just want to X-check this to avoid mixing things up (and they are very often mixed up in the net)
legacy417: Fair enough, I probably should have posted this in the discussion first anyway.
indiana stuff:
http://www.cs.ualberta.ca/~games/poker/man-machine/Press/edj_06-11-2007/ In Las Vegas in 2005, Vexbot adjusted its style in an attempt to exploit Laak's tendencies, but it didn't always make the best choices, Schaeffer says. "It was aggressive, it had a learning component, but it had an unstable learning component. This new program is more stable."
Vexbot winner here:
Another ref: http://www.canada.com/edmonton/edmontonjournal/news/story.html?id=85a006d7-54b7-42d4-a148-eab9f2c8cc28
Spears Can we publish this?
Vexbot is heads up limit adaptive pokerbot developed by the university of Alberta Computer Poker Research Group. It is described in "Game-Tree Search with Adaptation in Stochastic Imperfect-Information Games" by Darse Billings, Aaron Davidson, Terence Schauenberg, Neil Burch, Michael Bowling, Robert Holte, Jonathan Schaeffer, Duane Szafron. It uses the miximax algorithm to determine a best response to its opponent, and is similar to BRPlayer. It is one of the opponents available in Poker Academy. According to Poker Academy, you may find it quite a weak opponent to begin with, but it gets stronger after several hundred hands.
Vexbot won the gold medal at the Computer Olympiad held in Graz, Austria 2003. Sparbot, the game-theory based pokerbot, took silver medal in the same competition.
This press release from the University of Alberta made the following comments about the 2005 match in Las Vegas against Phil Laak: Laak easily beat Vexbot, ... partly because he played better, but also because he had far more luck that day -- as Laak himself readily admits. ... Vexbot adjusted its style in an attempt to exploit Laak's tendencies, but it didn't always make the best choices, Schaeffer says. "It was aggressive, it had a learning component, but it had an unstable learning component..."
Poker Academy promotional literature says: Vexbot ... is an AI system based completely on opponent modeling. This is a radical departure and very different from the original rule-based Loki. It has been successful thus far in proving that opponent modeling is critical when playing poker at a world class level. Vexbot forces an opponent to continually change strategies and adapt their play, as it will attempt to exploit any and all weakness or predictability it finds in their playing style. Players facing Vexbot find themselves asking "how can I win against myself?"
Indiana Looks good, go for it. I can then touch few non-content things, like cross references ...
