Hierarchical Deep Q-Network from imperfect demonstrations in Minecraft

Skrynnik A., Staroverov A., Aitygulov E., Aksenov K., Davydov V., Panov A.I.
Тип документа: Journal Article
Дата публикации: 2021-01-01
Название журнала: Cognitive Systems Research
Издатель: Elsevier
Квартиль: Q2
ISSN: 13890417
DOI:  10.1016/j.cogsys.2020.08.012
  • Artificial Intelligence
  • Software
  • Experimental and Cognitive Psychology
  • Cognitive Neuroscience
Краткое описание
We present Hierarchical Deep Q-Network (HDQfD) that took first place in the MineRL competition. HDQfD works on imperfect demonstrations and utilizes the hierarchical structure of expert trajectories. We introduce the procedure of extracting an effective sequence of meta-actions and subgoals from demonstration data. We present a structured task-dependent replay buffer and adaptive prioritizing technique that allow the HDQfD agent to gradually erase poor-quality expert data from the buffer. In this paper, we present the details of the HDQfD algorithm and give the experimental results in the Minecraft domain.
Пристатейные ссылки: 9
Цитируется в публикациях: 6
Метрики
Поделиться
Цитировать
ГОСТ |
Цитировать
ГОСТ Копировать
1. Skrynnik A. и др. Hierarchical Deep Q-Network from imperfect demonstrations in Minecraft // Cognitive Systems Research. 2021. Т. 65. С. 74–78.
RIS |
Цитировать
RIS Копировать

TY - JOUR

DO - 10.1016/j.cogsys.2020.08.012

UR - http://dx.doi.org/10.1016/j.cogsys.2020.08.012

TI - Hierarchical Deep Q-Network from imperfect demonstrations in Minecraft

T2 - Cognitive Systems Research

AU - Skrynnik, Alexey

AU - Staroverov, Aleksey

AU - Aitygulov, Ermek

AU - Aksenov, Kirill

AU - Davydov, Vasilii

AU - Panov, Aleksandr I.

PY - 2021

DA - 2021/01

PB - Elsevier BV

SP - 74-78

VL - 65

SN - 1389-0417

ER -

BibTex |
Цитировать
BibTex Копировать

@article{Skrynnik_2021,

doi = {10.1016/j.cogsys.2020.08.012},

url = {https://doi.org/10.1016%2Fj.cogsys.2020.08.012},

year = 2021,

month = {jan},

publisher = {Elsevier {BV}},

volume = {65},

pages = {74--78},

author = {Alexey Skrynnik and Aleksey Staroverov and Ermek Aitygulov and Kirill Aksenov and Vasilii Davydov and Aleksandr I. Panov},

title = {Hierarchical Deep Q-Network from imperfect demonstrations in Minecraft},

journal = {Cognitive Systems Research}

}

MLA
Цитировать
MLA Копировать
Skrynnik, Alexey et al. “Hierarchical Deep Q-Network from Imperfect Demonstrations in Minecraft.” Cognitive Systems Research 65 (2021): 74–78. Crossref. Web.