We propose a new framework for aiding a reinforcement learner by allowing it to relocate, or move, to a state it selects so as to decrease the number of steps it needs to take in order to develop an effective policy. The framework requires a minimal amount of human involvement or expertise and assumes a cost for each relocation. Several methods for taking advantage of the ability to relocate are proposed, and their effectiveness is tested in two commonly-used domains.

PDF, PS

In Prodeedings of the 19th International FLAIRS Conference (FLAIRS-2006), pp. 580-585, Melbourne Beach, FL, May 2006.

@InProceedings{mihalkova:flairs06, title={Using Active Relocation to Aid Reinforcement Learning}, author={Lilyana Mihalkova and Raymond Mooney}, booktitle={Prodeedings of the 19th International FLAIRS Conference (FLAIRS-2006)}, month={May}, address={Melbourne Beach, FL}, pages={580-585}, url="http://www.cs.utexas.edu/users/ai-lab?mihalkova:flairs06", year={2006} }