Open Access
Lecture Notes in Computer Science, volume 12854 LNAI, pages 224-235
Flexible Data Augmentation in Off-Policy Reinforcement Learning
Publication type: Book Chapter
Publication date: 2021-10-05
Journal:
Lecture Notes in Computer Science
Quartile SCImago
Q3
Quartile WOS
—
Impact factor: —
ISSN: 03029743, 16113349, 18612075, 18612083
Abstract
This paper explores an application of image augmentation in reinforcement learning tasks - a popular regularization technique in the computer vision area. The analysis is based on the model-free off-policy algorithms. As a regularization, we consider the augmentation of the frames that are sampled from the replay buffer of the model. Evaluated augmentation techniques are random changes in image contrast, random shifting, random cutting, and others. Research is done using the environments of the Atari games: Breakout, Space Invaders, Berzerk, Wizard of Wor, Demon Attack. Using augmentations allowed us to obtain results confirming the significant acceleration of the model’s algorithm convergence. We also proposed an adaptive mechanism for selecting the type of augmentation depending on the type of task being performed by the agent.
Metrics
Cite this
GOST |
RIS |
BibTex
Cite this
GOST
Copy
Rak A. et al. Flexible Data Augmentation in Off-Policy Reinforcement Learning // Lecture Notes in Computer Science. 2021. Vol. 12854 LNAI. pp. 224-235.
GOST all authors (up to 50)
Copy
Rak A., Skrynnik A., Panov A. I. Flexible Data Augmentation in Off-Policy Reinforcement Learning // Lecture Notes in Computer Science. 2021. Vol. 12854 LNAI. pp. 224-235.
Cite this
RIS
Copy
TY - GENERIC
DO - 10.1007/978-3-030-87986-0_20
UR - https://doi.org/10.1007%2F978-3-030-87986-0_20
TI - Flexible Data Augmentation in Off-Policy Reinforcement Learning
T2 - Lecture Notes in Computer Science
AU - Rak, Alexandra
AU - Skrynnik, Alexey
AU - Panov, Aleksandr I
PY - 2021
DA - 2021/10/05 00:00:00
PB - Springer Nature
SP - 224-235
VL - 12854 LNAI
SN - 0302-9743
SN - 1611-3349
SN - 1861-2075
SN - 1861-2083
ER -
Cite this
BibTex
Copy
@incollection{2021_Rak,
author = {Alexandra Rak and Alexey Skrynnik and Aleksandr I Panov},
title = {Flexible Data Augmentation in Off-Policy Reinforcement Learning},
publisher = {Springer Nature},
year = {2021},
volume = {12854 LNAI},
pages = {224--235},
month = {oct}
}
Profiles