Opponent modeling under partial observability in starcraft with deep convolutional encoder-decoders

Hyungu Kahng, Seoung Bum Kim

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

StarCraft, one of the most popular real-time strategy games, is a compelling environment for artificial intelligence research involving various tasks of both micro-level unit control and macro-level strategic decision making. In this study, we address an eminent problem of concern in macro-level decision making known as the “fog-of-war”, which rises from the partial observable nature of the game. Recovering information hidden under the fog can help capture advantageous high-level game dynamics, such as build orders, tactics and strategies of the opponent. Casted as a supervised learning problem, we propose a convolutional encoder-decoder architecture to predict potential counts and locations of the opponent’s units based on only partially visible and noisy state information. We visualize the model predictions on simplified grids to primarily evaluate the performance of our proposed method. Furthermore, we train an additional convolutional neural network classifier on the encoder-decoder outputs to predict the final winner of the game, as a means of demonstrating both effectiveness and applicability.

Original languageEnglish
Title of host publicationIntelligent Systems and Applications - Proceedings of the 2019 Intelligent Systems Conference IntelliSys Volume 1
EditorsYaxin Bi, Rahul Bhatia, Supriya Kapoor
PublisherSpringer Verlag
Pages751-759
Number of pages9
ISBN (Print)9783030295158
DOIs
Publication statusPublished - 2020 Jan 1
EventIntelligent Systems Conference, IntelliSys 2019 - London, United Kingdom
Duration: 2019 Sep 52019 Sep 6

Publication series

NameAdvances in Intelligent Systems and Computing
Volume1037
ISSN (Print)2194-5357
ISSN (Electronic)2194-5365

Conference

ConferenceIntelligent Systems Conference, IntelliSys 2019
CountryUnited Kingdom
CityLondon
Period19/9/519/9/6

Fingerprint

Observability
Fog
Macros
Decision making
Supervised learning
Artificial intelligence
Classifiers
Neural networks

Keywords

  • Convolutional neural networks
  • Fog-of-war
  • StarCraft

ASJC Scopus subject areas

  • Control and Systems Engineering
  • Computer Science(all)

Cite this

Kahng, H., & Kim, S. B. (2020). Opponent modeling under partial observability in starcraft with deep convolutional encoder-decoders. In Y. Bi, R. Bhatia, & S. Kapoor (Eds.), Intelligent Systems and Applications - Proceedings of the 2019 Intelligent Systems Conference IntelliSys Volume 1 (pp. 751-759). (Advances in Intelligent Systems and Computing; Vol. 1037). Springer Verlag. https://doi.org/10.1007/978-3-030-29516-5_56

Opponent modeling under partial observability in starcraft with deep convolutional encoder-decoders. / Kahng, Hyungu; Kim, Seoung Bum.

Intelligent Systems and Applications - Proceedings of the 2019 Intelligent Systems Conference IntelliSys Volume 1. ed. / Yaxin Bi; Rahul Bhatia; Supriya Kapoor. Springer Verlag, 2020. p. 751-759 (Advances in Intelligent Systems and Computing; Vol. 1037).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Kahng, H & Kim, SB 2020, Opponent modeling under partial observability in starcraft with deep convolutional encoder-decoders. in Y Bi, R Bhatia & S Kapoor (eds), Intelligent Systems and Applications - Proceedings of the 2019 Intelligent Systems Conference IntelliSys Volume 1. Advances in Intelligent Systems and Computing, vol. 1037, Springer Verlag, pp. 751-759, Intelligent Systems Conference, IntelliSys 2019, London, United Kingdom, 19/9/5. https://doi.org/10.1007/978-3-030-29516-5_56
Kahng H, Kim SB. Opponent modeling under partial observability in starcraft with deep convolutional encoder-decoders. In Bi Y, Bhatia R, Kapoor S, editors, Intelligent Systems and Applications - Proceedings of the 2019 Intelligent Systems Conference IntelliSys Volume 1. Springer Verlag. 2020. p. 751-759. (Advances in Intelligent Systems and Computing). https://doi.org/10.1007/978-3-030-29516-5_56
Kahng, Hyungu ; Kim, Seoung Bum. / Opponent modeling under partial observability in starcraft with deep convolutional encoder-decoders. Intelligent Systems and Applications - Proceedings of the 2019 Intelligent Systems Conference IntelliSys Volume 1. editor / Yaxin Bi ; Rahul Bhatia ; Supriya Kapoor. Springer Verlag, 2020. pp. 751-759 (Advances in Intelligent Systems and Computing).
@inproceedings{b9b87e1393ab4f1bbf3c515289277a6b,
title = "Opponent modeling under partial observability in starcraft with deep convolutional encoder-decoders",
abstract = "StarCraft, one of the most popular real-time strategy games, is a compelling environment for artificial intelligence research involving various tasks of both micro-level unit control and macro-level strategic decision making. In this study, we address an eminent problem of concern in macro-level decision making known as the “fog-of-war”, which rises from the partial observable nature of the game. Recovering information hidden under the fog can help capture advantageous high-level game dynamics, such as build orders, tactics and strategies of the opponent. Casted as a supervised learning problem, we propose a convolutional encoder-decoder architecture to predict potential counts and locations of the opponent’s units based on only partially visible and noisy state information. We visualize the model predictions on simplified grids to primarily evaluate the performance of our proposed method. Furthermore, we train an additional convolutional neural network classifier on the encoder-decoder outputs to predict the final winner of the game, as a means of demonstrating both effectiveness and applicability.",
keywords = "Convolutional neural networks, Fog-of-war, StarCraft",
author = "Hyungu Kahng and Kim, {Seoung Bum}",
year = "2020",
month = "1",
day = "1",
doi = "10.1007/978-3-030-29516-5_56",
language = "English",
isbn = "9783030295158",
series = "Advances in Intelligent Systems and Computing",
publisher = "Springer Verlag",
pages = "751--759",
editor = "Yaxin Bi and Rahul Bhatia and Supriya Kapoor",
booktitle = "Intelligent Systems and Applications - Proceedings of the 2019 Intelligent Systems Conference IntelliSys Volume 1",

}

TY - GEN

T1 - Opponent modeling under partial observability in starcraft with deep convolutional encoder-decoders

AU - Kahng, Hyungu

AU - Kim, Seoung Bum

PY - 2020/1/1

Y1 - 2020/1/1

N2 - StarCraft, one of the most popular real-time strategy games, is a compelling environment for artificial intelligence research involving various tasks of both micro-level unit control and macro-level strategic decision making. In this study, we address an eminent problem of concern in macro-level decision making known as the “fog-of-war”, which rises from the partial observable nature of the game. Recovering information hidden under the fog can help capture advantageous high-level game dynamics, such as build orders, tactics and strategies of the opponent. Casted as a supervised learning problem, we propose a convolutional encoder-decoder architecture to predict potential counts and locations of the opponent’s units based on only partially visible and noisy state information. We visualize the model predictions on simplified grids to primarily evaluate the performance of our proposed method. Furthermore, we train an additional convolutional neural network classifier on the encoder-decoder outputs to predict the final winner of the game, as a means of demonstrating both effectiveness and applicability.

AB - StarCraft, one of the most popular real-time strategy games, is a compelling environment for artificial intelligence research involving various tasks of both micro-level unit control and macro-level strategic decision making. In this study, we address an eminent problem of concern in macro-level decision making known as the “fog-of-war”, which rises from the partial observable nature of the game. Recovering information hidden under the fog can help capture advantageous high-level game dynamics, such as build orders, tactics and strategies of the opponent. Casted as a supervised learning problem, we propose a convolutional encoder-decoder architecture to predict potential counts and locations of the opponent’s units based on only partially visible and noisy state information. We visualize the model predictions on simplified grids to primarily evaluate the performance of our proposed method. Furthermore, we train an additional convolutional neural network classifier on the encoder-decoder outputs to predict the final winner of the game, as a means of demonstrating both effectiveness and applicability.

KW - Convolutional neural networks

KW - Fog-of-war

KW - StarCraft

UR - http://www.scopus.com/inward/record.url?scp=85072826582&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85072826582&partnerID=8YFLogxK

U2 - 10.1007/978-3-030-29516-5_56

DO - 10.1007/978-3-030-29516-5_56

M3 - Conference contribution

AN - SCOPUS:85072826582

SN - 9783030295158

T3 - Advances in Intelligent Systems and Computing

SP - 751

EP - 759

BT - Intelligent Systems and Applications - Proceedings of the 2019 Intelligent Systems Conference IntelliSys Volume 1

A2 - Bi, Yaxin

A2 - Bhatia, Rahul

A2 - Kapoor, Supriya

PB - Springer Verlag

ER -