MASAGE: Model-Agnostic Sequential and Adaptive Game Estimation

Yunian Pan; Guanze Peng; Juntao Chen; Quanyan Zhu

doi:10.1007/978-3-030-64793-3_20

MASAGE: Model-Agnostic Sequential and Adaptive Game Estimation

Yunian Pan, Guanze Peng, Juntao Chen, Quanyan Zhu

Electrical and Computer Engineering

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

Abstract

Zero-sum games have been used to model cybersecurity scenarios between an attacker and a defender. However, unknown and uncertain environments have made it difficult to rely on a prescribed zero-sum game to capture the interactions between the players. In this work, we aim to estimate and recover an unknown matrix game that encodes the uncertainties of nature and opponent based on the knowledge of historical games and the current observations of game outcomes. The proposed approach effectively transfers the past experiences that are encoded as expert games to estimate and inform future game plays. We formulate the game knowledge transfer and estimation problem as a sequential least-square problem. We characterize the structural properties of the problem and show that the non-convex problem has well-behaved gradient and Hessian under mild assumptions. We propose gradient-based methods to enable dynamic and adaptive estimation of the unknown game. A case study is used to corroborate the results and illustrate the behavior of the proposed algorithm.

Original language	English (US)
Title of host publication	Decision and Game Theory for Security - 11th International Conference, GameSec 2020, Proceedings
Editors	Quanyan Zhu, John S. Baras, Radha Poovendran, Juntao Chen
Publisher	Springer Science and Business Media Deutschland GmbH
Pages	365-384
Number of pages	20
ISBN (Print)	9783030647926
DOIs	https://doi.org/10.1007/978-3-030-64793-3_20
State	Published - 2020
Event	11th Conference on Decision and Game Theory for Security, GameSec 2020 - College Park, United States Duration: Oct 28 2020 → Oct 30 2020

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	12513 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	11th Conference on Decision and Game Theory for Security, GameSec 2020
Country/Territory	United States
City	College Park
Period	10/28/20 → 10/30/20

Keywords

Gradient-based methods
Least-square estimation
Neural networks
Security games
Sensitivity analysis
Zero-sum games

ASJC Scopus subject areas

Theoretical Computer Science
General Computer Science

Access to Document

10.1007/978-3-030-64793-3_20

Cite this

Pan, Y., Peng, G., Chen, J., & Zhu, Q. (2020). MASAGE: Model-Agnostic Sequential and Adaptive Game Estimation. In Q. Zhu, J. S. Baras, R. Poovendran, & J. Chen (Eds.), Decision and Game Theory for Security - 11th International Conference, GameSec 2020, Proceedings (pp. 365-384). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 12513 LNCS). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-3-030-64793-3_20

MASAGE: Model-Agnostic Sequential and Adaptive Game Estimation. / Pan, Yunian; Peng, Guanze; Chen, Juntao et al.
Decision and Game Theory for Security - 11th International Conference, GameSec 2020, Proceedings. ed. / Quanyan Zhu; John S. Baras; Radha Poovendran; Juntao Chen. Springer Science and Business Media Deutschland GmbH, 2020. p. 365-384 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 12513 LNCS).

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

Pan, Y, Peng, G, Chen, J & Zhu, Q 2020, MASAGE: Model-Agnostic Sequential and Adaptive Game Estimation. in Q Zhu, JS Baras, R Poovendran & J Chen (eds), Decision and Game Theory for Security - 11th International Conference, GameSec 2020, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 12513 LNCS, Springer Science and Business Media Deutschland GmbH, pp. 365-384, 11th Conference on Decision and Game Theory for Security, GameSec 2020, College Park, United States, 10/28/20. https://doi.org/10.1007/978-3-030-64793-3_20

Pan Y, Peng G, Chen J, Zhu Q. MASAGE: Model-Agnostic Sequential and Adaptive Game Estimation. In Zhu Q, Baras JS, Poovendran R, Chen J, editors, Decision and Game Theory for Security - 11th International Conference, GameSec 2020, Proceedings. Springer Science and Business Media Deutschland GmbH. 2020. p. 365-384. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-030-64793-3_20

Pan, Yunian ; Peng, Guanze ; Chen, Juntao et al. / MASAGE : Model-Agnostic Sequential and Adaptive Game Estimation. Decision and Game Theory for Security - 11th International Conference, GameSec 2020, Proceedings. editor / Quanyan Zhu ; John S. Baras ; Radha Poovendran ; Juntao Chen. Springer Science and Business Media Deutschland GmbH, 2020. pp. 365-384 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{0e7cc30143b74270942ffef402514051,

title = "MASAGE: Model-Agnostic Sequential and Adaptive Game Estimation",

abstract = "Zero-sum games have been used to model cybersecurity scenarios between an attacker and a defender. However, unknown and uncertain environments have made it difficult to rely on a prescribed zero-sum game to capture the interactions between the players. In this work, we aim to estimate and recover an unknown matrix game that encodes the uncertainties of nature and opponent based on the knowledge of historical games and the current observations of game outcomes. The proposed approach effectively transfers the past experiences that are encoded as expert games to estimate and inform future game plays. We formulate the game knowledge transfer and estimation problem as a sequential least-square problem. We characterize the structural properties of the problem and show that the non-convex problem has well-behaved gradient and Hessian under mild assumptions. We propose gradient-based methods to enable dynamic and adaptive estimation of the unknown game. A case study is used to corroborate the results and illustrate the behavior of the proposed algorithm.",

keywords = "Gradient-based methods, Least-square estimation, Neural networks, Security games, Sensitivity analysis, Zero-sum games",

author = "Yunian Pan and Guanze Peng and Juntao Chen and Quanyan Zhu",

note = "Funding Information: This research is partially supported by awards ECCS-1847056, CNS-1544782, CNS-2027884, and SES-1541164 from National Science of Foundation (NSF), and grant W911NF-19-1-0041 from Army Research Office (ARO). Publisher Copyright: {\textcopyright} 2020, Springer Nature Switzerland AG.; 11th Conference on Decision and Game Theory for Security, GameSec 2020 ; Conference date: 28-10-2020 Through 30-10-2020",

year = "2020",

doi = "10.1007/978-3-030-64793-3_20",

language = "English (US)",

isbn = "9783030647926",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer Science and Business Media Deutschland GmbH",

pages = "365--384",

editor = "Quanyan Zhu and Baras, {John S.} and Radha Poovendran and Juntao Chen",

booktitle = "Decision and Game Theory for Security - 11th International Conference, GameSec 2020, Proceedings",

address = "Germany",

}

TY - GEN

T1 - MASAGE

T2 - 11th Conference on Decision and Game Theory for Security, GameSec 2020

AU - Pan, Yunian

AU - Peng, Guanze

AU - Chen, Juntao

AU - Zhu, Quanyan

N1 - Funding Information: This research is partially supported by awards ECCS-1847056, CNS-1544782, CNS-2027884, and SES-1541164 from National Science of Foundation (NSF), and grant W911NF-19-1-0041 from Army Research Office (ARO). Publisher Copyright: © 2020, Springer Nature Switzerland AG.

PY - 2020

Y1 - 2020

N2 - Zero-sum games have been used to model cybersecurity scenarios between an attacker and a defender. However, unknown and uncertain environments have made it difficult to rely on a prescribed zero-sum game to capture the interactions between the players. In this work, we aim to estimate and recover an unknown matrix game that encodes the uncertainties of nature and opponent based on the knowledge of historical games and the current observations of game outcomes. The proposed approach effectively transfers the past experiences that are encoded as expert games to estimate and inform future game plays. We formulate the game knowledge transfer and estimation problem as a sequential least-square problem. We characterize the structural properties of the problem and show that the non-convex problem has well-behaved gradient and Hessian under mild assumptions. We propose gradient-based methods to enable dynamic and adaptive estimation of the unknown game. A case study is used to corroborate the results and illustrate the behavior of the proposed algorithm.

AB - Zero-sum games have been used to model cybersecurity scenarios between an attacker and a defender. However, unknown and uncertain environments have made it difficult to rely on a prescribed zero-sum game to capture the interactions between the players. In this work, we aim to estimate and recover an unknown matrix game that encodes the uncertainties of nature and opponent based on the knowledge of historical games and the current observations of game outcomes. The proposed approach effectively transfers the past experiences that are encoded as expert games to estimate and inform future game plays. We formulate the game knowledge transfer and estimation problem as a sequential least-square problem. We characterize the structural properties of the problem and show that the non-convex problem has well-behaved gradient and Hessian under mild assumptions. We propose gradient-based methods to enable dynamic and adaptive estimation of the unknown game. A case study is used to corroborate the results and illustrate the behavior of the proposed algorithm.

KW - Gradient-based methods

KW - Least-square estimation

KW - Neural networks

KW - Security games

KW - Sensitivity analysis

KW - Zero-sum games

UR - http://www.scopus.com/inward/record.url?scp=85098287728&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85098287728&partnerID=8YFLogxK

U2 - 10.1007/978-3-030-64793-3_20

DO - 10.1007/978-3-030-64793-3_20

M3 - Conference contribution

AN - SCOPUS:85098287728

SN - 9783030647926

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

SP - 365

EP - 384

BT - Decision and Game Theory for Security - 11th International Conference, GameSec 2020, Proceedings

A2 - Zhu, Quanyan

A2 - Baras, John S.

A2 - Poovendran, Radha

A2 - Chen, Juntao

PB - Springer Science and Business Media Deutschland GmbH

Y2 - 28 October 2020 through 30 October 2020

ER -

MASAGE: Model-Agnostic Sequential and Adaptive Game Estimation

Abstract

Publication series

Conference

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this