25 October 1993, WeH 4601, 3:00

		 The Parti-Game algorithm for online
	      variable resolution reinforcement learning

			     Andrew Moore

Notice: This talk is a technical, speculative version of a more
civilized talk which will be given at the AI Seminar on November 2nd.

This talk describes the technical details of a new algorithm called
Parti-Game. In high dimensional continuous state spaces it is
essential that learning does not explore or plan over state space
uniformly.  Parti-game maintains a decision-tree partitioning of
state-space and applies techniques from game-theory and computational
geometry to efficiently and adaptively concentrate high resolution
only on critical areas.  The current version of the algorithm is
designed to find feasible solutions to high dimensional problems.
Future versions will be designed to find a solution that optimizes a
real-valued criterion.  Many simulated problems have been tested, and
will be described and demonstrated.