Print Email Facebook Twitter Risk Aversion and Guided Exploration in Safety-Constrained Reinforcement Learning Title Risk Aversion and Guided Exploration in Safety-Constrained Reinforcement Learning Author Yang, Q. (TU Delft Algorithmics) Contributor Spaan, M.T.J. (promotor) Tindemans, Simon H. (copromotor) Degree granting institution Delft University of Technology Date 2023-06-23 Abstract In traditional reinforcement learning (RL) problems, agents can explore environments to learn optimal policies through trials and errors that are sometimes unsafe. However, unsafe interactions with environments are unacceptable in many safety-critical problems, for instance in robot navigation tasks. Even though RL agents can be trained in simulators, there are many real-world problems without simulators of sufficient fidelity. Constructing safe exploration algorithms for dangerous environments is challenging because we have to optimize policies under the premise of safety. In general, safety is still an open problem that hinders the wider application of RL. Subject Reinforcement Leaning (RL)constrained optimizationquantile regressiontaskagnostic exploration To reference this document use: https://doi.org/10.4233/uuid:ca5a81c2-f895-4638-bce5-1423a5943381 ISBN 978-94-6384-458-1 Part of collection Institutional Repository Document type doctoral thesis Rights © 2023 Q. Yang Files PDF Dissertation_QisongYang_1_.pdf 28.41 MB PDF PhD_propositions_qisong.pdf 111.17 KB Close viewer /islandora/object/uuid:ca5a81c2-f895-4638-bce5-1423a5943381/datastream/OBJ1/view