dc.contributor.author | Mannion, Patrick | |
dc.contributor.author | Devlin, Sam | |
dc.contributor.author | Karl, Mannion | |
dc.contributor.author | Duggan, Jim | |
dc.date.accessioned | 2018-12-19T16:32:02Z | |
dc.date.available | 2018-12-19T16:32:02Z | |
dc.date.copyright | 2017-05 | |
dc.date.issued | 2017-05 | |
dc.identifier.uri | https://research.thea.ie/handle/20.500.12065/2391 | |
dc.description.abstract | Reward shaping is a well-established family of techniques
that have been successfully used to improve the performance
and learning speed of Reinforcement Learning agents in singleobjective
problems. Here we extend the guarantees of Potential-
Based Reward Shaping (PBRS) by providing theoretical
proof that PBRS does not alter the true Pareto front in
MORL domains. We also contribute the rst empirical studies
of the e ect of PBRS in MORL problems. | en_US |
dc.format | Pdf | en_US |
dc.language.iso | en | en_US |
dc.rights | Attribution-NonCommercial-NoDerivs 3.0 Ireland | * |
dc.rights.uri | http://creativecommons.org/licenses/by-nc-nd/3.0/ie/ | * |
dc.subject | Multi-Objective | en_US |
dc.subject | Reinforcement Learning | en_US |
dc.subject | Reward Shaping | en_US |
dc.title | Potential-Based Reward Shaping Preserves Pareto Optimal Policies | en_US |
dc.description.peerreview | yes | en_US |
dc.rights.access | Copyright | en_US |
dc.subject.department | Department of Computer Science & Applied Physics | en_US |