An object-oriented representation for efficient reinforcement learning.

Carlos Diuk, Andre Cohen, Michael L. Littman

Browse the full ICML paper archive.