Towards sample-efficient policy learning with DAC-ML
Mostra el registre complet Registre parcial de l'ítem
- dc.contributor.author Verschure, Paul F. M. J.
- dc.contributor.author Freire, Ismael T.
- dc.contributor.author Amil, Adrián F.
- dc.contributor.author Vouloutsi, Vasiliki
- dc.date.accessioned 2023-03-01T13:51:46Z
- dc.date.available 2023-03-01T13:51:46Z
- dc.date.issued 2021
- dc.description.abstract The sample-inefficiency problem in Artificial Intelligence refers to the inability of current Deep Reinforcement Learning models to optimize action policies within a small number of episodes. Recent studies have tried to overcome this limitation by adding memory systems and architectural biases to improve learning speed, such as in Episodic Reinforcement Learning. However, despite achieving incremental improvements, their performance is still not comparable to how humans learn behavioral policies. In this paper, we capitalize on the design principles of the Distributed Adaptive Control (DAC) theory of mind and brain to build a novel cognitive architecture (DAC-ML) that, by incorporating a hippocampus-inspired sequential memory system, can rapidly converge to effective action policies that maximize reward acquisition in a challenging foraging task.
- dc.description.sponsorship This research received funding from the H2020-EU project HR-Recycler, ID: 820742.
- dc.format.mimetype application/pdf
- dc.identifier.citation Freire IT, Amil AF, Vouloutsi V, Verschure PFMJ. Towards sample-efficient policy learning with DAC-ML. Procedia Comput Sci. 2021;190:256–62. DOI: 10.1016/j.procs.2021.06.035
- dc.identifier.doi http://dx.doi.org/10.1016/j.procs.2021.06.035
- dc.identifier.issn 1877-0509
- dc.identifier.uri http://hdl.handle.net/10230/55998
- dc.language.iso eng
- dc.publisher Elsevier
- dc.relation.ispartof Procedia Computer Science. 2021;190:256–62.
- dc.relation.projectID info:eu-repo/grantAgreement/EC/H2020/820742
- dc.rights © 2021 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0)
- dc.rights.accessRights info:eu-repo/semantics/openAccess
- dc.rights.uri https://creativecommons.org/licenses/by-nc-nd/4.0
- dc.subject.keyword cognitive architecture
- dc.subject.keyword sample-inefficiency problem
- dc.subject.keyword sequence learning
- dc.subject.keyword reinforcement learning
- dc.subject.keyword distributed adaptive control
- dc.title Towards sample-efficient policy learning with DAC-ML
- dc.type info:eu-repo/semantics/article
- dc.type.version info:eu-repo/semantics/publishedVersion