Predictive Representations to Improve Generalization in Reinforcement
Learning, Eddie J. Rafols, Mark B. Ring, Richard S.
Sutton, Brian Tanner; Proceedings
of the 19th
International Joint Conference on Artificial Intelligence, 2005.
predictive representations hypothesis holds that particularly good
generalization will result from representing the state of the world in
terms of predictions about possible future experience. This hypothesis
has been a central motivation behind recent research in, for example,
PSRs and TD networks. In this paper we present the first explicit
investigation of this hypothesis. We show in a reinforcement-learning
example (a grid-world navigation task) that a predictive representation
in tabular form can learn much faster than both the tabular
explicit-state representation and a tabular history-based method.