On robustness of CMAB algorithms: Experimental approach

Antonín Komenda; Alexander Shleyfman; Carmel Domshlak

doi:10.1007/978-3-319-14923-3

On robustness of CMAB algorithms: Experimental approach

Antonín Komenda, Alexander Shleyfman, Carmel Domshlak

Technion-Israel Institute of Technology

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

1 Scopus citations

Abstract

In this paper, we experimentally analyze robustness of two state-of-the- art algorithms NMC and LSI for online planning with combinatorial actions in various setups of Real-Time and Turn-Taking Strategy games.

In online planning with a team of cooperative agents, a straightforward model for decision making which actions the agents should execute can be represented as the problem of Combinatorial Multi-Armed Bandit. Similarly to the most prominent approaches for online planning with polynomial number of possible actions, state-of-the-art algorithms for online planning with exponential number of actions are based on Monte-Carlo sampling. However, without a proper selection of the appropriate subset of actions these techniques cannot be used. The most recent algorithms tackling this problem utilize an assumption of linearity with respect to the combinations of the actions.

Original language	English
Title of host publication	Computer Games - 3rd Workshop on Computer Games, CGW 2014 Held in Conjunction with the 21st European Conference on Artificial Intelligence, ECAI 2014, Revised Selected Papers
Editors	Tristan Cazenave, Yngvi Björnsson, Mark H.M. Winands
Publisher	Springer Verlag
Pages	16-28
Number of pages	13
ISBN (Electronic)	9783319149226
DOIs	https://doi.org/10.1007/978-3-319-14923-3
State	Published - 2014
Externally published	Yes
Event	3rd Workshop on Computer Games, CGW 2014 held in Conjunction with the 21st European Conference on Artificial Intelligence, ECAI 2014 - Prague, Czech Republic Duration: 18 Aug 2014 → 18 Aug 2014

Publication series

Name	Communications in Computer and Information Science
Volume	504
ISSN (Print)	1865-0929

Conference

Conference	3rd Workshop on Computer Games, CGW 2014 held in Conjunction with the 21st European Conference on Artificial Intelligence, ECAI 2014
Country/Territory	Czech Republic
City	Prague
Period	18/08/14 → 18/08/14

Bibliographical note

Publisher Copyright:
© Springer International Publishing Switzerland 2014.

Access to Document

10.1007/978-3-319-14923-3

Cite this

Komenda, A., Shleyfman, A., & Domshlak, C. (2014). On robustness of CMAB algorithms: Experimental approach. In T. Cazenave, Y. Björnsson, & M. H. M. Winands (Eds.), Computer Games - 3rd Workshop on Computer Games, CGW 2014 Held in Conjunction with the 21st European Conference on Artificial Intelligence, ECAI 2014, Revised Selected Papers (pp. 16-28). (Communications in Computer and Information Science; Vol. 504). Springer Verlag. https://doi.org/10.1007/978-3-319-14923-3

Komenda, Antonín ; Shleyfman, Alexander ; Domshlak, Carmel. / On robustness of CMAB algorithms : Experimental approach. Computer Games - 3rd Workshop on Computer Games, CGW 2014 Held in Conjunction with the 21st European Conference on Artificial Intelligence, ECAI 2014, Revised Selected Papers. editor / Tristan Cazenave ; Yngvi Björnsson ; Mark H.M. Winands. Springer Verlag, 2014. pp. 16-28 (Communications in Computer and Information Science).

@inproceedings{8a19f7013cf5490dba4ebe8a7e569b68,

title = "On robustness of CMAB algorithms: Experimental approach",

abstract = "In this paper, we experimentally analyze robustness of two state-of-the- art algorithms NMC and LSI for online planning with combinatorial actions in various setups of Real-Time and Turn-Taking Strategy games.In online planning with a team of cooperative agents, a straightforward model for decision making which actions the agents should execute can be represented as the problem of Combinatorial Multi-Armed Bandit. Similarly to the most prominent approaches for online planning with polynomial number of possible actions, state-of-the-art algorithms for online planning with exponential number of actions are based on Monte-Carlo sampling. However, without a proper selection of the appropriate subset of actions these techniques cannot be used. The most recent algorithms tackling this problem utilize an assumption of linearity with respect to the combinations of the actions.",

author = "Anton{\'i}n Komenda and Alexander Shleyfman and Carmel Domshlak",

note = "Publisher Copyright: {\textcopyright} Springer International Publishing Switzerland 2014.; 3rd Workshop on Computer Games, CGW 2014 held in Conjunction with the 21st European Conference on Artificial Intelligence, ECAI 2014 ; Conference date: 18-08-2014 Through 18-08-2014",

year = "2014",

doi = "10.1007/978-3-319-14923-3",

language = "אנגלית",

series = "Communications in Computer and Information Science",

publisher = "Springer Verlag",

pages = "16--28",

editor = "Tristan Cazenave and Yngvi Bj{\"o}rnsson and Winands, {Mark H.M.}",

booktitle = "Computer Games - 3rd Workshop on Computer Games, CGW 2014 Held in Conjunction with the 21st European Conference on Artificial Intelligence, ECAI 2014, Revised Selected Papers",

address = "גרמניה",

}

Komenda, A, Shleyfman, A & Domshlak, C 2014, On robustness of CMAB algorithms: Experimental approach. in T Cazenave, Y Björnsson & MHM Winands (eds), Computer Games - 3rd Workshop on Computer Games, CGW 2014 Held in Conjunction with the 21st European Conference on Artificial Intelligence, ECAI 2014, Revised Selected Papers. Communications in Computer and Information Science, vol. 504, Springer Verlag, pp. 16-28, 3rd Workshop on Computer Games, CGW 2014 held in Conjunction with the 21st European Conference on Artificial Intelligence, ECAI 2014, Prague, Czech Republic, 18/08/14. https://doi.org/10.1007/978-3-319-14923-3

On robustness of CMAB algorithms: Experimental approach. / Komenda, Antonín; Shleyfman, Alexander; Domshlak, Carmel.
Computer Games - 3rd Workshop on Computer Games, CGW 2014 Held in Conjunction with the 21st European Conference on Artificial Intelligence, ECAI 2014, Revised Selected Papers. ed. / Tristan Cazenave; Yngvi Björnsson; Mark H.M. Winands. Springer Verlag, 2014. p. 16-28 (Communications in Computer and Information Science; Vol. 504).

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

TY - GEN

T1 - On robustness of CMAB algorithms

T2 - 3rd Workshop on Computer Games, CGW 2014 held in Conjunction with the 21st European Conference on Artificial Intelligence, ECAI 2014

AU - Komenda, Antonín

AU - Shleyfman, Alexander

AU - Domshlak, Carmel

PY - 2014

Y1 - 2014

N2 - In this paper, we experimentally analyze robustness of two state-of-the- art algorithms NMC and LSI for online planning with combinatorial actions in various setups of Real-Time and Turn-Taking Strategy games.In online planning with a team of cooperative agents, a straightforward model for decision making which actions the agents should execute can be represented as the problem of Combinatorial Multi-Armed Bandit. Similarly to the most prominent approaches for online planning with polynomial number of possible actions, state-of-the-art algorithms for online planning with exponential number of actions are based on Monte-Carlo sampling. However, without a proper selection of the appropriate subset of actions these techniques cannot be used. The most recent algorithms tackling this problem utilize an assumption of linearity with respect to the combinations of the actions.

AB - In this paper, we experimentally analyze robustness of two state-of-the- art algorithms NMC and LSI for online planning with combinatorial actions in various setups of Real-Time and Turn-Taking Strategy games.In online planning with a team of cooperative agents, a straightforward model for decision making which actions the agents should execute can be represented as the problem of Combinatorial Multi-Armed Bandit. Similarly to the most prominent approaches for online planning with polynomial number of possible actions, state-of-the-art algorithms for online planning with exponential number of actions are based on Monte-Carlo sampling. However, without a proper selection of the appropriate subset of actions these techniques cannot be used. The most recent algorithms tackling this problem utilize an assumption of linearity with respect to the combinations of the actions.

UR - http://www.scopus.com/inward/record.url?scp=84919449614&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-14923-3

DO - 10.1007/978-3-319-14923-3

M3 - ???researchoutput.researchoutputtypes.contributiontobookanthology.conference???

AN - SCOPUS:84919449614

T3 - Communications in Computer and Information Science

SP - 16

EP - 28

BT - Computer Games - 3rd Workshop on Computer Games, CGW 2014 Held in Conjunction with the 21st European Conference on Artificial Intelligence, ECAI 2014, Revised Selected Papers

A2 - Cazenave, Tristan

A2 - Björnsson, Yngvi

A2 - Winands, Mark H.M.

PB - Springer Verlag

Y2 - 18 August 2014 through 18 August 2014

ER -

Komenda A, Shleyfman A, Domshlak C. On robustness of CMAB algorithms: Experimental approach. In Cazenave T, Björnsson Y, Winands MHM, editors, Computer Games - 3rd Workshop on Computer Games, CGW 2014 Held in Conjunction with the 21st European Conference on Artificial Intelligence, ECAI 2014, Revised Selected Papers. Springer Verlag. 2014. p. 16-28. (Communications in Computer and Information Science). doi: 10.1007/978-3-319-14923-3

On robustness of CMAB algorithms: Experimental approach

Abstract

Publication series

Conference

Bibliographical note

Access to Document

Other files and links

Fingerprint

Cite this