Cargando…
Experience Replay Using Transition Sequences
Experience replay is one of the most commonly used approaches to improve the sample efficiency of reinforcement learning algorithms. In this work, we propose an approach to select and replay sequences of transitions in order to accelerate the learning of a reinforcement learning agent in an off-poli...
Autores principales: | Karimpanal, Thommen George, Bouffanais, Roland |
---|---|
Formato: | Online Artículo Texto |
Lenguaje: | English |
Publicado: |
Frontiers Media S.A.
2018
|
Materias: | |
Acceso en línea: | https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6022201/ https://www.ncbi.nlm.nih.gov/pubmed/29977200 http://dx.doi.org/10.3389/fnbot.2018.00032 |
Ejemplares similares
-
Beyond Bio-Inspired Robotics: How Multi-Robot Systems Can Support Research on Collective Animal Behavior
por: Horsevad, Nikolaj, et al.
Publicado: (2022) -
Balancing Collective Exploration and Exploitation in Multi-Agent and Multi-Robot Systems: A Review
por: Kwa , Hian Lee, et al.
Publicado: (2022) -
Adaptivity: a path towards general swarm intelligence?
por: Kwa, Hian Lee, et al.
Publicado: (2023) -
Robust Understanding of Robot-Directed Speech Commands Using Sequence to Sequence With Noise Injection
por: Tada, Yuuki, et al.
Publicado: (2020) -
Attraction, Dynamics, and Phase Transitions in Fire Ant Tower-Building
por: Nave, Gary K., et al.
Publicado: (2020)