Learning Policies for Markov Decision Processes From Data

DOI10.1109/TAC.2018.2866455MaRDI QIDQ5223736zbMATH OpenOpenAlexWikidataFDO

Authors Manjesh Kumar Hanawal, Hao Liu, Henghui Zhu, Ioannis Ch. Paschalidis

Publication date 18 July 2019

Published in IEEE Transactions on Automatic Control (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/1701.05954

Mathematics Subject Classification ID

Markov processes: estimation; hidden Markov models (62M05) Markov and semi-Markov decision processes (90C40) Stochastic learning and adaptive control (93E35)

Abstract: We consider the problem of learning a policy for a Markov decision process consistent with data captured on the state-actions pairs followed by the policy. We assume that the policy belongs to a class of parameterized policies which are defined using features associated with the state-action pairs. The features are known a priori, however, only an unknown subset of them could be relevant. The policy parameters that correspond to an observed target policy are recovered using $el{l}_1$-regularized logistic regression that best fits the observed state-action samples. We establish bounds on the difference between the average reward of the estimated and the original policy (regret) in terms of the generalization error and the ergodic coefficient of the underlying Markov chain. To that end, we combine sample complexity theory and sensitivity analysis of the stationary distribution of Markov chains. Our analysis suggests that to achieve regret within order $O(sqrtepsilon)$, it suffices to use training sample size on the order of $Omega(logncdotpoly(1/epsilon))$, where $n$ is the number of the features. We demonstrate the effectiveness of our method on a synthetic robot navigation example.

Cited in

(6)

• \(L^\ast\)-based learning of Markov decision processes (extended version)
• Learning algorithms for Markov decision processes
• Learning Variable-Length Markov Models of Behavior
• Bagging strategies for learning planning policies
• On a probabilistic approach to synthesize control policies from example datasets
• Learning parametric policies and transition probability models of Markov decision processes from data

This page was built for publication: Learning Policies for Markov Decision Processes From Data

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5223736)