Knowledge based planning in games

This paper explores the issue of planning in two-person games using approximately learned knowledge which is in the form of macros. A planning technique called Knowledge Enabled Planning, which consists of generating new plans by instantiating and composing the macros and testing them against the opponent's counter-plans, is introduced using a program called LEBL that learns and plans in two-person games. Some empirical results of testing LEBL in king and pawn endings in chess are presented along with a complexity analysis of the planning algorithm. This research was partially supported by the National Science Foundation under Contract Number IRI-8740522, and the Defense Advanced Research Projects Agency under Contract Number N00014-85-K-0116. The views and conclusions contained in this document are those of the author and should not be interpreted as representing the official policies, either expressed or implied, of the National Science Foundation, the Defense Advanced Research Projects Agency, or the U.S. Government.