Zobrazeno 1 - 6
of 6
pro vyhledávání: '"Nicolas Fontbonne"'
Publikováno v:
PLoS ONE, Vol 17, Iss 4 (2022)
This paper focuses on a class of reinforcement learning problems where significant events are rare and limited to a single positive reward per episode. A typical example is that of an agent who has to choose a partner to cooperate with, while a large
Externí odkaz:
https://doaj.org/article/f989505a58d749ac87745c014cc6d1b7
Cooperative Co-evolution and Adaptive Team Composition for a Multi-rover Resource Allocation Problem
Publikováno v:
Lecture Notes in Computer Science ISBN: 9783031020551
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_________::a496f209b5c308fe0da8e9a987ddf789
https://doi.org/10.1007/978-3-031-02056-8_12
https://doi.org/10.1007/978-3-031-02056-8_12
Autor:
Nicolas Bredeche, Nicolas Fontbonne
Publikováno v:
Philosophical Transactions of the Royal Society B: Biological Sciences
Philosophical Transactions of the Royal Society B: Biological Sciences, Royal Society, The, 2022, 377 (1843), pp.20200309. ⟨10.1098/rstb.2020.0309⟩
Philosophical Transactions of the Royal Society B: Biological Sciences, Royal Society, The, 2022, 377 (1843), pp.20200309. ⟨10.1098/rstb.2020.0309⟩
In this paper, we present an implementation of social learning for swarm robotics. We consider social learning as a distributed online reinforcement learning method applied to a collective of robots where sensing, acting and coordination are performe
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_dedup___::5e872bc9dea580d93221c2698074b222
https://doi.org/10.1098/rstb.2020.0309
https://doi.org/10.1098/rstb.2020.0309
Reinforcement Learning with Rare Significant Events: Direct Policy Search vs. Gradient Policy Search
Publikováno v:
GECCO 21 : Proceedings of the 2021 Genetic and Evolutionary Computation Conference Companion
Genetic and Evolutionary Computation Conference Companion
Genetic and Evolutionary Computation Conference Companion, 2021, Lille (en ligne), France
GECCO Companion
Genetic and Evolutionary Computation Conference Companion
Genetic and Evolutionary Computation Conference Companion, 2021, Lille (en ligne), France
GECCO Companion
This paper shows that the CMAES direct policy search method fares significantly better than PPO gradient policy search for a reinforcement learning task where significant events are rare.
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_dedup___::9ae067ba30302bb63a3af7e5851a6389
https://hal.sorbonne-universite.fr/hal-03315728
https://hal.sorbonne-universite.fr/hal-03315728
This paper focuses on a class of reinforcement learning problems where significant events are rare and limited to a single positive reward per episode. A typical example is that of an agent who has to choose a partner to cooperate with, while a large
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_dedup___::19d2d50d8e3da2ebb4d2eaec9dcf2063
Publikováno v:
2020 IEEE Congress on Evolutionary Computation (CEC)
IEEE Congress on Evolutionary Computation
IEEE Congress on Evolutionary Computation, 2020, Glasgow (virtual), United Kingdom. ⟨10.1109/CEC48606.2020.9185697⟩
CEC
IEEE Congress on Evolutionary Computation
IEEE Congress on Evolutionary Computation, 2020, Glasgow (virtual), United Kingdom. ⟨10.1109/CEC48606.2020.9185697⟩
CEC
International audience; This paper presents a new algorithm for distributed on-line evolutionary learning in swarm robotics. The challenge we address is to cope with the limited computation and communication capabilities of low cost robots, which are
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_dedup___::bf56793dd961db60562c72838a6831f5
https://hal.sorbonne-universite.fr/hal-03175237/document
https://hal.sorbonne-universite.fr/hal-03175237/document