Sciweavers

AAAI
1998

Finding Optimal Strategies for Imperfect Information Games

13 years 5 months ago
Finding Optimal Strategies for Imperfect Information Games
Weexaminethree heuristic algorithms for gameswith imperfect information: Monte-carlo sampling, and two newalgorithms wecall vector minimaxingand payoffreduction minimaxing. Wecomparethese algorithms theoretically and experimentally, using both simple gametrees and a large database of problemsfrom the game of Bridge. Our experiments show that the new algorithms both out-perform Monte-carlo sampling, with the superiority of payoff-reduction minimaxing being especially marked. Onthe Bridge problemset, for example, Monte-carlo sampling only solves 66% of the problems, whereas payoff-reduction minimaxing solves over 95%.This level of performance was evengoodenoughto allowus to discover five errors in the expert text used to generate the test database.
Ian Frank, David A. Basin, Hitoshi Matsubara
Added 01 Nov 2010
Updated 01 Nov 2010
Type Conference
Year 1998
Where AAAI
Authors Ian Frank, David A. Basin, Hitoshi Matsubara
Comments (0)