Learning human-like behaviors using NeuroEvolution with statistical penalties

Luong Huu Phuc; Kanazawa Naoto; Kokolo Ikeda

doi:10.1109/CIG.2017.8080437

Learning human-like behaviors using NeuroEvolution with statistical penalties

Luong Huu Phuc, Kanazawa Naoto, Kokolo Ikeda

Research output: Chapter in Book/Report/Conference proceeding › Conference Paper › Research › peer-review

3 Citations (Scopus)

Abstract

In game artificial intelligence (AI), two common directions for developing non-human computer players are strong AI and human-like AI. Human-like AI aims at making computer agents behave like humans. In this direction, NeuroEvolution (NE), which is a combination of an artificial neural network (ANN) and an evolutionary algorithm (EA), had been frequently used to a make computer agent to behave like a human. Our research introduces a novel approach to create human-like computer agents in a platform game Super Mario Bros. (SMB) - we called it a 2D action game in this research. The approach utilizes statistical penalties to evaluate candidates created by NE algorithm. The penalties help in reducing mechanical actions of computer agents based on human data statistics, and the effects of statistical penalties are analyzed by asking human subjects to rate the human-likeness of agents. Experiments show that our method improves the human-likeness in the behavior of a computer agent.

Original language	English
Title of host publication	2017 IEEE Conference on Computational Intelligence and Games (CIG)
Editors	Clare Bates Congdon, Michael Buro
Place of Publication	Piscataway NJ USA
Publisher	IEEE, Institute of Electrical and Electronics Engineers
Pages	207-214
Number of pages	8
ISBN (Electronic)	9781538632338, 9781538632321
ISBN (Print)	9781538632345
DOIs	https://doi.org/10.1109/CIG.2017.8080437
Publication status	Published - 2017
Externally published	Yes
Event	IEEE Symposium on Computational Intelligence and Games 2017 - New York, United States of America Duration: 22 Aug 2017 → 25 Aug 2017 http://www.cig2017.com

Publication series

Name	2017 IEEE Conference on Computational Intelligence and Games, CIG 2017
ISSN (Electronic)	2325-4289

Conference

Conference	IEEE Symposium on Computational Intelligence and Games 2017
Abbreviated title	CIG 2017
Country/Territory	United States of America
City	New York
Period	22/08/17 → 25/08/17
Internet address	http://www.cig2017.com

Access to Document

10.1109/CIG.2017.8080437

Cite this

Phuc, L. H., Naoto, K., & Ikeda, K. (2017). Learning human-like behaviors using NeuroEvolution with statistical penalties. In C. B. Congdon, & M. Buro (Eds.), 2017 IEEE Conference on Computational Intelligence and Games (CIG) (pp. 207-214). (2017 IEEE Conference on Computational Intelligence and Games, CIG 2017). IEEE, Institute of Electrical and Electronics Engineers. https://doi.org/10.1109/CIG.2017.8080437

Phuc, Luong Huu ; Naoto, Kanazawa ; Ikeda, Kokolo. / Learning human-like behaviors using NeuroEvolution with statistical penalties. 2017 IEEE Conference on Computational Intelligence and Games (CIG). editor / Clare Bates Congdon ; Michael Buro. Piscataway NJ USA : IEEE, Institute of Electrical and Electronics Engineers, 2017. pp. 207-214 (2017 IEEE Conference on Computational Intelligence and Games, CIG 2017).

@inproceedings{cc831d2ecd2045b0ba1315b1184a6a07,

title = "Learning human-like behaviors using NeuroEvolution with statistical penalties",

abstract = "In game artificial intelligence (AI), two common directions for developing non-human computer players are strong AI and human-like AI. Human-like AI aims at making computer agents behave like humans. In this direction, NeuroEvolution (NE), which is a combination of an artificial neural network (ANN) and an evolutionary algorithm (EA), had been frequently used to a make computer agent to behave like a human. Our research introduces a novel approach to create human-like computer agents in a platform game Super Mario Bros. (SMB) - we called it a 2D action game in this research. The approach utilizes statistical penalties to evaluate candidates created by NE algorithm. The penalties help in reducing mechanical actions of computer agents based on human data statistics, and the effects of statistical penalties are analyzed by asking human subjects to rate the human-likeness of agents. Experiments show that our method improves the human-likeness in the behavior of a computer agent.",

author = "Phuc, {Luong Huu} and Kanazawa Naoto and Kokolo Ikeda",

note = "Publisher Copyright: {\textcopyright} 2017 IEEE. Copyright: Copyright 2018 Elsevier B.V., All rights reserved.; IEEE Symposium on Computational Intelligence and Games 2017, CIG 2017 ; Conference date: 22-08-2017 Through 25-08-2017",

year = "2017",

doi = "10.1109/CIG.2017.8080437",

language = "English",

isbn = "9781538632345",

series = "2017 IEEE Conference on Computational Intelligence and Games, CIG 2017",

publisher = "IEEE, Institute of Electrical and Electronics Engineers",

pages = "207--214",

editor = "Congdon, {Clare Bates} and Michael Buro",

booktitle = "2017 IEEE Conference on Computational Intelligence and Games (CIG)",

address = "United States of America",

url = "http://www.cig2017.com",

}

Phuc, LH, Naoto, K & Ikeda, K 2017, Learning human-like behaviors using NeuroEvolution with statistical penalties. in CB Congdon & M Buro (eds), 2017 IEEE Conference on Computational Intelligence and Games (CIG). 2017 IEEE Conference on Computational Intelligence and Games, CIG 2017, IEEE, Institute of Electrical and Electronics Engineers, Piscataway NJ USA, pp. 207-214, IEEE Symposium on Computational Intelligence and Games 2017, New York, New York, United States of America, 22/08/17. https://doi.org/10.1109/CIG.2017.8080437

Learning human-like behaviors using NeuroEvolution with statistical penalties. / Phuc, Luong Huu; Naoto, Kanazawa; Ikeda, Kokolo.

2017 IEEE Conference on Computational Intelligence and Games (CIG). ed. / Clare Bates Congdon; Michael Buro. Piscataway NJ USA : IEEE, Institute of Electrical and Electronics Engineers, 2017. p. 207-214 (2017 IEEE Conference on Computational Intelligence and Games, CIG 2017).

Research output: Chapter in Book/Report/Conference proceeding › Conference Paper › Research › peer-review

TY - GEN

T1 - Learning human-like behaviors using NeuroEvolution with statistical penalties

AU - Phuc, Luong Huu

AU - Naoto, Kanazawa

AU - Ikeda, Kokolo

PY - 2017

Y1 - 2017

N2 - In game artificial intelligence (AI), two common directions for developing non-human computer players are strong AI and human-like AI. Human-like AI aims at making computer agents behave like humans. In this direction, NeuroEvolution (NE), which is a combination of an artificial neural network (ANN) and an evolutionary algorithm (EA), had been frequently used to a make computer agent to behave like a human. Our research introduces a novel approach to create human-like computer agents in a platform game Super Mario Bros. (SMB) - we called it a 2D action game in this research. The approach utilizes statistical penalties to evaluate candidates created by NE algorithm. The penalties help in reducing mechanical actions of computer agents based on human data statistics, and the effects of statistical penalties are analyzed by asking human subjects to rate the human-likeness of agents. Experiments show that our method improves the human-likeness in the behavior of a computer agent.

AB - In game artificial intelligence (AI), two common directions for developing non-human computer players are strong AI and human-like AI. Human-like AI aims at making computer agents behave like humans. In this direction, NeuroEvolution (NE), which is a combination of an artificial neural network (ANN) and an evolutionary algorithm (EA), had been frequently used to a make computer agent to behave like a human. Our research introduces a novel approach to create human-like computer agents in a platform game Super Mario Bros. (SMB) - we called it a 2D action game in this research. The approach utilizes statistical penalties to evaluate candidates created by NE algorithm. The penalties help in reducing mechanical actions of computer agents based on human data statistics, and the effects of statistical penalties are analyzed by asking human subjects to rate the human-likeness of agents. Experiments show that our method improves the human-likeness in the behavior of a computer agent.

UR - http://www.scopus.com/inward/record.url?scp=85039982103&partnerID=8YFLogxK

U2 - 10.1109/CIG.2017.8080437

DO - 10.1109/CIG.2017.8080437

M3 - Conference Paper

AN - SCOPUS:85039982103

SN - 9781538632345

T3 - 2017 IEEE Conference on Computational Intelligence and Games, CIG 2017

SP - 207

EP - 214

BT - 2017 IEEE Conference on Computational Intelligence and Games (CIG)

A2 - Congdon, Clare Bates

A2 - Buro, Michael

PB - IEEE, Institute of Electrical and Electronics Engineers

CY - Piscataway NJ USA

T2 - IEEE Symposium on Computational Intelligence and Games 2017

Y2 - 22 August 2017 through 25 August 2017

ER -

Phuc LH, Naoto K, Ikeda K. Learning human-like behaviors using NeuroEvolution with statistical penalties. In Congdon CB, Buro M, editors, 2017 IEEE Conference on Computational Intelligence and Games (CIG). Piscataway NJ USA: IEEE, Institute of Electrical and Electronics Engineers. 2017. p. 207-214. (2017 IEEE Conference on Computational Intelligence and Games, CIG 2017). https://doi.org/10.1109/CIG.2017.8080437

Learning human-like behaviors using NeuroEvolution with statistical penalties

Abstract

Publication series

Conference

Access to Document

Other files and links

Cite this