Memory-based Reinforcement Learning: Efficient Computation with Prioritized Sweeping - Robotics Institute Carnegie Mellon University

Memory-based Reinforcement Learning: Efficient Computation with Prioritized Sweeping

Andrew Moore and C. G. Atkeson
Conference Paper, Proceedings of (NeurIPS) Neural Information Processing Systems, pp. 263 - 270, November, 1992

Abstract

We present a new algorithm, Prioritized Sweeping, for efficient prediction and control of stochastic Markov systems. Incremental learning methods such as Temporal Differencing and Q-Iearning have fast real time performance. Classical methods are slower, but more accurate, because they make full use of the observations. Prioritized Sweeping aims for the best of both worlds. It uses all previous experiences both to prioritize important dynamic programming sweeps and to guide the exploration of statespace. We compare Prioritized Sweeping with other reinforcement learning schemes for a number of different stochastic optimal control problems. It successfully solves large state-space real time problems with which other methods have difficulty.

BibTeX

@conference{Moore-1992-15876,
author = {Andrew Moore and C. G. Atkeson},
title = {Memory-based Reinforcement Learning: Efficient Computation with Prioritized Sweeping},
booktitle = {Proceedings of (NeurIPS) Neural Information Processing Systems},
year = {1992},
month = {November},
editor = {S. J. Hanson, J. D Cowan, and C. L. Giles},
pages = {263 - 270},
publisher = {Morgan Kaufmann},
}