Monte-Carlo Sampling in Games with Incomplete Information:
empirical investigation and analysis
Ian Frank, David Basin, Hitoshi Matsubara
Presented at the Workshop on Game Tree Search, Princeton, NJ,
1997. Book version now in press (to appear in March 1999).
The version on these pages is a slightly earlier ETL Technical Report,
ETL-TR-97-18
Abstract:
We investigate Monte-carlo sampling in games with incomplete
information. We show that for very simple game trees the chance of
finding the optimal strategy with Monte-carlo sampling rapidly
approaches zero as the number of moves in the game increases. We
explain this sub-optimality by identifying the different kinds of
errors that can arise and by analysing their interplay. We also
relate our test results to real games, suggesting why the error
rates observed in practice may not be so high.
|