Abstract
We study upper and lower bounds on the sample-complexity of learning near-optimal behaviour in finite-state discounted Markov Decision Processes (mdps). We prove a new bound for a modified version of Upper Confidence Reinforcement Learning (ucrl) with only cubic dependence on the horizon. The bound is unimprovable in all parameters except the size of the state/action space, where it depends linearly on the number of non-zero transition probabilities. The lower bound strengthens previous work by being both more general (it applies to all policies) and tighter. The upper and lower bounds match up to logarithmic factors provided the transition matrix is not too dense.
This is a preview of subscription content, log in via an institution.
Buying options
Tax calculation will be finalised at checkout
Purchases are for personal use only
Learn about institutional subscriptionsPreview
Unable to display preview. Download preview PDF.
References
Auer, P., Jaksch, T., Ortner, R.: Near-optimal regret bounds for reinforcement learning. J. Mach. Learn. Res. 99, 1563–1600 (2010)
Azar, M., Munos, R., Kappen, B.: On the sample complexity of reinforcement learning with a generative model. In: Proceedings of the 29th International Conference on Machine Learning. ACM, New York (2012)
Auer, P., Ortner, R.: Logarithmic online regret bounds for undiscounted reinforcement learning. In: Advances in Neural Information Processing Systems 19, pp. 49–56. MIT Press (2007)
Auer, P.: Upper confidence reinforcement learning. Unpublished, keynote at European Workshop of Reinforcement Learning (2011)
Chung, F., Lu, L.: Concentration inequalities and martingale inequalities a survey. Internet Mathematics 3, 1 (2006)
Kakade, S.: On The Sample Complexity of Reinforcement Learning. PhD thesis, University College London (2003)
Lattimore, T., Hutter, M.: PAC bounds for discounted MDPs. Technical report (2012), http://arxiv.org/abs/1202.3890
Lai, T., Robbins, H.: Asymptotically efficient adaptive allocation rules. Advances in Applied Mathematics 6(1), 4–22 (1985)
Mannor, S., Tsitsiklis, J.: The sample complexity of exploration in the multi-armed bandit problem. J. Mach. Learn. Res. 5, 623–648 (2004)
Strehl, A., Littman, M.: A theoretical analysis of model-based interval estimation. In: Proceedings of the 22nd International Conference on Machine Learning, ICML 2005, pp. 856–863 (2005)
Strehl, A., Littman, M.: An analysis of model-based interval estimation for Markov decision processes. Journal of Computer and System Sciences 74(8), 1309–1331 (2008)
Strehl, A., Li, L., Littman, M.: Reinforcement learning in finite MDPs: PAC analysis. J. Mach. Learn. Res. 10, 2413–2444 (2009)
Strehl, A., Li, L., Wiewiorac, E., Langford, J., Littman, M.: PAC model-free reinforcement learning. In: Proceedings of the 23rd International Conference on Machine Learning, ICML 2006, pp. 881–888. ACM, New York (2006)
Sobel, M.: The variance of discounted Markov decision processes. Journal of Applied Probability 19(4), 794–802 (1982)
Szita, I., Szepesvári, C.: Model-based reinforcement learning with nearly tight exploration complexity bounds. In: Proceedings of the 27th International Conference on Machine Learning, pp. 1031–1038. ACM, New York (2010)
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2012 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Lattimore, T., Hutter, M. (2012). PAC Bounds for Discounted MDPs. In: Bshouty, N.H., Stoltz, G., Vayatis, N., Zeugmann, T. (eds) Algorithmic Learning Theory. ALT 2012. Lecture Notes in Computer Science(), vol 7568. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-642-34106-9_26
Download citation
DOI: https://doi.org/10.1007/978-3-642-34106-9_26
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-642-34105-2
Online ISBN: 978-3-642-34106-9
eBook Packages: Computer ScienceComputer Science (R0)