Monte-Carlo simulation balancing in practice

Shih Chieh Huang, Rémi Coulom, Shun-Shii Lin

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

28 Citations (Scopus)

Abstract

Simulation balancing is a new technique to tune parameters of a playout policy for a Monte-Carlo game-playing program. So far, this algorithm had only been tested in a very artificial setting: it was limited to 5x5 and 6x6 Go, and required a stronger external program that served as a supervisor. In this paper, the effectiveness of simulation balancing is demonstrated in a more realistic setting. A state-of-the-art program, Erica, learned an improved playout policy on the 9x9 board, without requiring any external expert to provide position evaluations. The evaluations were collected by letting the program analyze positions by itself. The previous version of Erica learned pattern weights with the minorization-maximization algorithm. Thanks to simulation balancing, its playing strength was improved from a winning rate of 69% to 78% against Fuego 0.4.

Original languageEnglish
Title of host publicationComputers and Games - 7th International Conference, CG 2010, Revised Selected Papers
PublisherSpringer Verlag
Pages81-92
Number of pages12
ISBN (Print)3642179274, 9783642179273
DOIs
Publication statusPublished - 2011 Jan 1

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume6515 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Fingerprint

Balancing
Monte Carlo Simulation
Supervisory personnel
Simulation
Evaluation
Game
Monte Carlo simulation
Policy

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

Huang, S. C., Coulom, R., & Lin, S-S. (2011). Monte-Carlo simulation balancing in practice. In Computers and Games - 7th International Conference, CG 2010, Revised Selected Papers (pp. 81-92). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 6515 LNCS). Springer Verlag. https://doi.org/10.1007/978-3-642-17928-0_8

Monte-Carlo simulation balancing in practice. / Huang, Shih Chieh; Coulom, Rémi; Lin, Shun-Shii.

Computers and Games - 7th International Conference, CG 2010, Revised Selected Papers. Springer Verlag, 2011. p. 81-92 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 6515 LNCS).

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

Huang, SC, Coulom, R & Lin, S-S 2011, Monte-Carlo simulation balancing in practice. in Computers and Games - 7th International Conference, CG 2010, Revised Selected Papers. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 6515 LNCS, Springer Verlag, pp. 81-92. https://doi.org/10.1007/978-3-642-17928-0_8
Huang SC, Coulom R, Lin S-S. Monte-Carlo simulation balancing in practice. In Computers and Games - 7th International Conference, CG 2010, Revised Selected Papers. Springer Verlag. 2011. p. 81-92. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-642-17928-0_8
Huang, Shih Chieh ; Coulom, Rémi ; Lin, Shun-Shii. / Monte-Carlo simulation balancing in practice. Computers and Games - 7th International Conference, CG 2010, Revised Selected Papers. Springer Verlag, 2011. pp. 81-92 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{ce7ff4edfe9d4811a7126d6b0562947b,
title = "Monte-Carlo simulation balancing in practice",
abstract = "Simulation balancing is a new technique to tune parameters of a playout policy for a Monte-Carlo game-playing program. So far, this algorithm had only been tested in a very artificial setting: it was limited to 5x5 and 6x6 Go, and required a stronger external program that served as a supervisor. In this paper, the effectiveness of simulation balancing is demonstrated in a more realistic setting. A state-of-the-art program, Erica, learned an improved playout policy on the 9x9 board, without requiring any external expert to provide position evaluations. The evaluations were collected by letting the program analyze positions by itself. The previous version of Erica learned pattern weights with the minorization-maximization algorithm. Thanks to simulation balancing, its playing strength was improved from a winning rate of 69{\%} to 78{\%} against Fuego 0.4.",
author = "Huang, {Shih Chieh} and R{\'e}mi Coulom and Shun-Shii Lin",
year = "2011",
month = "1",
day = "1",
doi = "10.1007/978-3-642-17928-0_8",
language = "English",
isbn = "3642179274",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "81--92",
booktitle = "Computers and Games - 7th International Conference, CG 2010, Revised Selected Papers",

}

TY - GEN

T1 - Monte-Carlo simulation balancing in practice

AU - Huang, Shih Chieh

AU - Coulom, Rémi

AU - Lin, Shun-Shii

PY - 2011/1/1

Y1 - 2011/1/1

N2 - Simulation balancing is a new technique to tune parameters of a playout policy for a Monte-Carlo game-playing program. So far, this algorithm had only been tested in a very artificial setting: it was limited to 5x5 and 6x6 Go, and required a stronger external program that served as a supervisor. In this paper, the effectiveness of simulation balancing is demonstrated in a more realistic setting. A state-of-the-art program, Erica, learned an improved playout policy on the 9x9 board, without requiring any external expert to provide position evaluations. The evaluations were collected by letting the program analyze positions by itself. The previous version of Erica learned pattern weights with the minorization-maximization algorithm. Thanks to simulation balancing, its playing strength was improved from a winning rate of 69% to 78% against Fuego 0.4.

AB - Simulation balancing is a new technique to tune parameters of a playout policy for a Monte-Carlo game-playing program. So far, this algorithm had only been tested in a very artificial setting: it was limited to 5x5 and 6x6 Go, and required a stronger external program that served as a supervisor. In this paper, the effectiveness of simulation balancing is demonstrated in a more realistic setting. A state-of-the-art program, Erica, learned an improved playout policy on the 9x9 board, without requiring any external expert to provide position evaluations. The evaluations were collected by letting the program analyze positions by itself. The previous version of Erica learned pattern weights with the minorization-maximization algorithm. Thanks to simulation balancing, its playing strength was improved from a winning rate of 69% to 78% against Fuego 0.4.

UR - http://www.scopus.com/inward/record.url?scp=79952027913&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=79952027913&partnerID=8YFLogxK

U2 - 10.1007/978-3-642-17928-0_8

DO - 10.1007/978-3-642-17928-0_8

M3 - Conference contribution

AN - SCOPUS:79952027913

SN - 3642179274

SN - 9783642179273

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 81

EP - 92

BT - Computers and Games - 7th International Conference, CG 2010, Revised Selected Papers

PB - Springer Verlag

ER -