SvgAI - Training artificial intelligent agent to use SVG editor

Anh H. Dang; Wataru Kameyama

doi:10.23919/ICACT.2018.8323672

SvgAI - Training artificial intelligent agent to use SVG editor

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Abstract

Deep reinforcement learning has been successfully used to train artificial intelligent (AI) agents to outperform humans in many tasks as well as to enhance the capability in robotic automation. In this paper, we propose a framework to train an AI agent to use scalable vector graphic (SVG) editor to draw SVG images. Hence, the objective of this AI agent is to draw SVG images that are similar as much as possible to their target raster images. We find that it is crucial to distinguish the action space into two sets and apply a different exploration policy on each set during the training process. Evaluations show that our proposed dual-exploration policy greatly stabilizes the training process and increases the accuracy of the AI agent. SVG images produced by the proposed AI agent also have superior quality compared to popular raster-to-SVG conversion software.

Original language	English
Title of host publication	IEEE 20th International Conference on Advanced Communication Technology
Subtitle of host publication	Opening New Era of Intelligent Things, ICACT 2018
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	132-138
Number of pages	7
Volume	2018-February
ISBN (Electronic)	9791188428007
DOIs	https://doi.org/10.23919/ICACT.2018.8323672
Publication status	Published - 2018 Mar 23
Event	20th IEEE International Conference on Advanced Communication Technology, ICACT 2018 - Chuncheon, Korea, Republic of Duration: 2018 Feb 11 → 2018 Feb 14

Other

Other	20th IEEE International Conference on Advanced Communication Technology, ICACT 2018
Country/Territory	Korea, Republic of
City	Chuncheon
Period	18/2/11 → 18/2/14

Keywords

Exploration Policy
Q-Learning
Reinforcement Learning
SVG
SvgAI

ASJC Scopus subject areas

Electrical and Electronic Engineering

Access to Document

10.23919/ICACT.2018.8323672

Cite this

SvgAI - Training artificial intelligent agent to use SVG editor. / Dang, Anh H.; Kameyama, Wataru.

IEEE 20th International Conference on Advanced Communication Technology: Opening New Era of Intelligent Things, ICACT 2018. Vol. 2018-February Institute of Electrical and Electronics Engineers Inc., 2018. p. 132-138.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Dang, AH & Kameyama, W 2018, SvgAI - Training artificial intelligent agent to use SVG editor. in IEEE 20th International Conference on Advanced Communication Technology: Opening New Era of Intelligent Things, ICACT 2018. vol. 2018-February, Institute of Electrical and Electronics Engineers Inc., pp. 132-138, 20th IEEE International Conference on Advanced Communication Technology, ICACT 2018, Chuncheon, Korea, Republic of, 18/2/11. https://doi.org/10.23919/ICACT.2018.8323672

@inproceedings{847b40031f894c629e2de12ce6c1f350,

title = "SvgAI - Training artificial intelligent agent to use SVG editor",

abstract = "Deep reinforcement learning has been successfully used to train artificial intelligent (AI) agents to outperform humans in many tasks as well as to enhance the capability in robotic automation. In this paper, we propose a framework to train an AI agent to use scalable vector graphic (SVG) editor to draw SVG images. Hence, the objective of this AI agent is to draw SVG images that are similar as much as possible to their target raster images. We find that it is crucial to distinguish the action space into two sets and apply a different exploration policy on each set during the training process. Evaluations show that our proposed dual-exploration policy greatly stabilizes the training process and increases the accuracy of the AI agent. SVG images produced by the proposed AI agent also have superior quality compared to popular raster-to-SVG conversion software.",

keywords = "Exploration Policy, Q-Learning, Reinforcement Learning, SVG, SvgAI",

author = "Dang, {Anh H.} and Wataru Kameyama",

year = "2018",

month = mar,

day = "23",

doi = "10.23919/ICACT.2018.8323672",

language = "English",

volume = "2018-February",

pages = "132--138",

booktitle = "IEEE 20th International Conference on Advanced Communication Technology",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

note = "20th IEEE International Conference on Advanced Communication Technology, ICACT 2018 ; Conference date: 11-02-2018 Through 14-02-2018",

}

TY - GEN

T1 - SvgAI - Training artificial intelligent agent to use SVG editor

AU - Dang, Anh H.

AU - Kameyama, Wataru

PY - 2018/3/23

Y1 - 2018/3/23

N2 - Deep reinforcement learning has been successfully used to train artificial intelligent (AI) agents to outperform humans in many tasks as well as to enhance the capability in robotic automation. In this paper, we propose a framework to train an AI agent to use scalable vector graphic (SVG) editor to draw SVG images. Hence, the objective of this AI agent is to draw SVG images that are similar as much as possible to their target raster images. We find that it is crucial to distinguish the action space into two sets and apply a different exploration policy on each set during the training process. Evaluations show that our proposed dual-exploration policy greatly stabilizes the training process and increases the accuracy of the AI agent. SVG images produced by the proposed AI agent also have superior quality compared to popular raster-to-SVG conversion software.

AB - Deep reinforcement learning has been successfully used to train artificial intelligent (AI) agents to outperform humans in many tasks as well as to enhance the capability in robotic automation. In this paper, we propose a framework to train an AI agent to use scalable vector graphic (SVG) editor to draw SVG images. Hence, the objective of this AI agent is to draw SVG images that are similar as much as possible to their target raster images. We find that it is crucial to distinguish the action space into two sets and apply a different exploration policy on each set during the training process. Evaluations show that our proposed dual-exploration policy greatly stabilizes the training process and increases the accuracy of the AI agent. SVG images produced by the proposed AI agent also have superior quality compared to popular raster-to-SVG conversion software.

KW - Exploration Policy

KW - Q-Learning

KW - Reinforcement Learning

KW - SVG

KW - SvgAI

UR - http://www.scopus.com/inward/record.url?scp=85046708720&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85046708720&partnerID=8YFLogxK

U2 - 10.23919/ICACT.2018.8323672

DO - 10.23919/ICACT.2018.8323672

M3 - Conference contribution

AN - SCOPUS:85046708720

VL - 2018-February

SP - 132

EP - 138

BT - IEEE 20th International Conference on Advanced Communication Technology

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 20th IEEE International Conference on Advanced Communication Technology, ICACT 2018

Y2 - 11 February 2018 through 14 February 2018

ER -

SvgAI - Training artificial intelligent agent to use SVG editor

Abstract

Other

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this