Abstract
In reinforcement learning the interaction between the agent and the environment generally takes place on a fixed time scale, which means that the control interval is set to a fixed time step. In order to determine a suitable fixed time scale one has to trade off accuracy in control against learning complexity. In this paper, we present an alternative approach that enables the agent to learn a control policy by using multiple time scales simultaneously. Instead of preselecting a fixed time scale, there are several time scales available during learning and the agent can select the appropriate time scale depending on the system state. The different time scales are multiples of a finest time scale which is denoted as the primitive time scale. Actions on a coarser time scale consist of several identical actions on the primitive time scale and are called multistep actions (MSAs). The special structure of these actions is efficiently exploited in our recently proposed MSA-Q-learning algorithm. In this paper, we use the MSAs to learn a control policy for a thermostat control problem. Our algorithm yields a fast and highly accurate control policy; in contrast, the standard Q-learning algorithms without MSAs fails to learn any useful control policy for this problem.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
Similar content being viewed by others
References
T. G. Dietterich. Hierarchical reinforcement learning with the MAXQ value function decomposition. Journal of Artificial Intelligence Research, 13:227–303, 2000.
S. Pareigis. Adaptive choice of grid and time in reinforcement learning. NIPS, volume 10. MIT Press, 1998.
R. E. Parr. Hierarchical Control and Learning for Markov Decision Processes. PhD thesis, University of California, Berkeley, CA, 1998.
M. Riedmiller. High quality thermostat control by reinforcement learning — a case study. In Proceedings of the Conald Workshop 1998, CMU, 1998.
M. Riedmiller. Concepts and facilities of a neural reinforcement learning control architecture for technical process control. Journal of Neural Computing and Application, 8:323–338, 2000.
R. Schoknecht and M. Riedmiller. Speeding-up reinforcement learning with multistep actions. ICANN, LNCS 2415, pages 813–818, 2002. Springer.
R. S. Sutton. Generalization in reinforcement learning: Successful examples using sparse coarse coding. NIPS, volume 8, pages 1038–1044. MIT Press, 1996.
R. S. Sutton, D. Precup, and S. Singh. Between mdps and semi-mdps: A framework for temporal abstraction in reinforcement learning. AI, 112:181–211, 1999.
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2003 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Schoknecht, R., Riedmiller, M. (2003). Learning to Control at Multiple Time Scales. In: Kaynak, O., Alpaydin, E., Oja, E., Xu, L. (eds) Artificial Neural Networks and Neural Information Processing — ICANN/ICONIP 2003. ICANN ICONIP 2003 2003. Lecture Notes in Computer Science, vol 2714. Springer, Berlin, Heidelberg. https://doi.org/10.1007/3-540-44989-2_57
Download citation
DOI: https://doi.org/10.1007/3-540-44989-2_57
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-40408-8
Online ISBN: 978-3-540-44989-8
eBook Packages: Springer Book Archive