Combining deep reinforcement learning with prior knowledge and reasoning
ACM SIGAPP Applied Computing Review · 26.07.2018 · 18 (2) · 33-45
DOI: 10.1145/3243064.3243067 ↗Kurz gefasst
Fachartikel in der Zeitschrift ACM SIGAPP Applied Computing Review, online erschienen am 26.07.2018. Verfasst von Nicolas Bougie und 2 weiteren Autor:innen. Die Arbeit wurde bisher 12-mal zitiert (Crossref) und nennt 36 Literaturangaben.
Abstract (Originalsprache)
Recent improvements in deep reinforcement learning have allowed to solve problems in many 2D domains such as Atari games. However, in complex 3D environments, numerous learning episodes are required which may be too time consuming or even impossible especially in real-world scenarios. We present a new architecture to combine external knowledge and deep reinforcement learning using only visual input. A key concept of our system is augmenting image input by adding environment feature information and combining two sources of decision. We evaluate the performances of our method in 3D partially-observable environments from the Microsoft Malmo platform. Experimental evaluation exhibits higher performance and faster learning compared to a single reinforcement learning model.
Zitieren
APA
Bougie, N., Cheng, L. K., & Ichise, R. (2018). Combining deep reinforcement learning with prior knowledge and reasoning. ACM SIGAPP Applied Computing Review, 18(2), 33-45. https://doi.org/10.1145/3243064.3243067
Vancouver
Bougie N, Cheng LK, Ichise R. Combining deep reinforcement learning with prior knowledge and reasoning. ACM SIGAPP Applied Computing Review. 2018;18(2):33-45. doi:10.1145/3243064.3243067
BibTeX
@article{Bougie2018,
author = {Bougie, Nicolas and Cheng, Li Kai and Ichise, Ryutaro},
title = {Combining deep reinforcement learning with prior knowledge and reasoning},
journal = {ACM SIGAPP Applied Computing Review},
year = {2018},
volume = {18},
number = {2},
pages = {33--45},
doi = {10.1145/3243064.3243067}
}
EUVIMED