Safe Policy Improvement Approaches on Discrete Markov Decision Processes

Philipp Scholl, Philipp Scholl, Felix Dietrich, Clemens Otte, Steffen Udluft

2022

Abstract

Safe Policy Improvement (SPI) aims at provable guarantees that a learned policy is at least approximately as good as a given baseline policy. Building on SPI with Soft Baseline Bootstrapping (Soft-SPIBB) by Nadjahi et al., we identify theoretical issues in their approach, provide a corrected theory, and derive a new algorithm that is provably safe on finite Markov Decision Processes (MDP). Additionally, we provide a heuristic algorithm that exhibits the best performance among many state of the art SPI algorithms on two different benchmarks. Furthermore, we introduce a taxonomy of SPI algorithms and empirically show an interesting property of two classes of SPI algorithms: while the mean performance of algorithms that incorporate the uncertainty as a penalty on the action-value is higher, actively restricting the set of policies more consistently produces good policies and is, thus, safer.

Paper Citation

in Harvard Style

Scholl P., Dietrich F., Otte C. and Udluft S. (2022). Safe Policy Improvement Approaches on Discrete Markov Decision Processes. In Proceedings of the 14th International Conference on Agents and Artificial Intelligence - Volume 2: ICAART, ISBN 978-989-758-547-0, pages 142-151. DOI: 10.5220/0010786600003116

in Bibtex Style

@conference{icaart22,
author={Philipp Scholl and Felix Dietrich and Clemens Otte and Steffen Udluft},
title={Safe Policy Improvement Approaches on Discrete Markov Decision Processes},
booktitle={Proceedings of the 14th International Conference on Agents and Artificial Intelligence - Volume 2: ICAART,},
year={2022},
pages={142-151},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0010786600003116},
isbn={978-989-758-547-0},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 14th International Conference on Agents and Artificial Intelligence - Volume 2: ICAART,
TI - Safe Policy Improvement Approaches on Discrete Markov Decision Processes
SN - 978-989-758-547-0
AU - Scholl P.
AU - Dietrich F.
AU - Otte C.
AU - Udluft S.
PY - 2022
SP - 142
EP - 151
DO - 10.5220/0010786600003116