Using active relocation to aid reinforcement learning
Lilyana Simeonova Mihalkova, Raymond J. Mooney · 2006
We propose a new framework for aiding a reinforcement learner by allowing it to relocate, or move, to a state it se-lects 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 as-sumes 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.