Home > Research > Publications & Outputs > Stochastic Fictitious Play using Particle Filte...
View graph of relations

Stochastic Fictitious Play using Particle Filters to update the beliefs of opponents strategies

Research output: Contribution in Book/Report/Proceedings - With ISBN/ISSNConference contribution/Paperpeer-review

Published

Standard

Stochastic Fictitious Play using Particle Filters to update the beliefs of opponents strategies. / Smyrnakis, Michalis; Leslie, David S.
Optimization in Multi-Agent Systems workshop OptMas, 17th International Conference on Autonomous Agents and Multiagent Systems AAMAS 2008.. ed. / Lin Padgham; David Parkes; Jörg Müller. INESC-ID, 2008.

Research output: Contribution in Book/Report/Proceedings - With ISBN/ISSNConference contribution/Paperpeer-review

Harvard

Smyrnakis, M & Leslie, DS 2008, Stochastic Fictitious Play using Particle Filters to update the beliefs of opponents strategies. in L Padgham, D Parkes & J Müller (eds), Optimization in Multi-Agent Systems workshop OptMas, 17th International Conference on Autonomous Agents and Multiagent Systems AAMAS 2008.. INESC-ID. <http://users.ecs.soton.ac.uk/sdr/optmas/accepted/paper8.pdf>

APA

Smyrnakis, M., & Leslie, D. S. (2008). Stochastic Fictitious Play using Particle Filters to update the beliefs of opponents strategies. In L. Padgham, D. Parkes, & J. Müller (Eds.), Optimization in Multi-Agent Systems workshop OptMas, 17th International Conference on Autonomous Agents and Multiagent Systems AAMAS 2008. INESC-ID. http://users.ecs.soton.ac.uk/sdr/optmas/accepted/paper8.pdf

Vancouver

Smyrnakis M, Leslie DS. Stochastic Fictitious Play using Particle Filters to update the beliefs of opponents strategies. In Padgham L, Parkes D, Müller J, editors, Optimization in Multi-Agent Systems workshop OptMas, 17th International Conference on Autonomous Agents and Multiagent Systems AAMAS 2008.. INESC-ID. 2008

Author

Smyrnakis, Michalis ; Leslie, David S. / Stochastic Fictitious Play using Particle Filters to update the beliefs of opponents strategies. Optimization in Multi-Agent Systems workshop OptMas, 17th International Conference on Autonomous Agents and Multiagent Systems AAMAS 2008.. editor / Lin Padgham ; David Parkes ; Jörg Müller. INESC-ID, 2008.

Bibtex

@inproceedings{cbd7020ea2914ea0aab15c41a88fe486,
title = "Stochastic Fictitious Play using Particle Filters to update the beliefs of opponents strategies",
abstract = "Distributed optimization can be formulated as an n player coordination game. One of the most common learning techniques in game theory is fictitious play and its variations. However fictitious play is founded on an implicit assumptionthat opponents{\textquoteright} strategies are stationary. In this paper we present a new variation of fictitious play in which players predict opponents{\textquoteright} strategy using a particle filter algorithm. This allows us to use a more realistic model of opponent strategy. We used pre-specified opponents{\textquoteright} strategies to examine if our algorithm can efficiently track the strategies. Furthermore we have used these experiments to examine the impact of different values of our algorithm parameters on the results of strategy tracking. We then compared the results of the proposed algorithm with those of stochastic and dynamic fictitious play in a potential game and two climbing hill games, one with two players and the other with three players. Our algorithm converges more quickly to the optimum thanboth the competitor algorithms. Hence by placing a greater computational demand on the individual agents, less communication is required between the agents.",
author = "Michalis Smyrnakis and Leslie, {David S.}",
note = "BAE SYSTEMS and ESPSRC funding for the ALADDIN project (EP/C548051/1)",
year = "2008",
language = "English",
isbn = "9780981738109",
editor = "Lin Padgham and David Parkes and J{\"o}rg M{\"u}ller",
booktitle = "Optimization in Multi-Agent Systems workshop OptMas, 17th International Conference on Autonomous Agents and Multiagent Systems AAMAS 2008.",
publisher = "INESC-ID",

}

RIS

TY - GEN

T1 - Stochastic Fictitious Play using Particle Filters to update the beliefs of opponents strategies

AU - Smyrnakis, Michalis

AU - Leslie, David S.

N1 - BAE SYSTEMS and ESPSRC funding for the ALADDIN project (EP/C548051/1)

PY - 2008

Y1 - 2008

N2 - Distributed optimization can be formulated as an n player coordination game. One of the most common learning techniques in game theory is fictitious play and its variations. However fictitious play is founded on an implicit assumptionthat opponents’ strategies are stationary. In this paper we present a new variation of fictitious play in which players predict opponents’ strategy using a particle filter algorithm. This allows us to use a more realistic model of opponent strategy. We used pre-specified opponents’ strategies to examine if our algorithm can efficiently track the strategies. Furthermore we have used these experiments to examine the impact of different values of our algorithm parameters on the results of strategy tracking. We then compared the results of the proposed algorithm with those of stochastic and dynamic fictitious play in a potential game and two climbing hill games, one with two players and the other with three players. Our algorithm converges more quickly to the optimum thanboth the competitor algorithms. Hence by placing a greater computational demand on the individual agents, less communication is required between the agents.

AB - Distributed optimization can be formulated as an n player coordination game. One of the most common learning techniques in game theory is fictitious play and its variations. However fictitious play is founded on an implicit assumptionthat opponents’ strategies are stationary. In this paper we present a new variation of fictitious play in which players predict opponents’ strategy using a particle filter algorithm. This allows us to use a more realistic model of opponent strategy. We used pre-specified opponents’ strategies to examine if our algorithm can efficiently track the strategies. Furthermore we have used these experiments to examine the impact of different values of our algorithm parameters on the results of strategy tracking. We then compared the results of the proposed algorithm with those of stochastic and dynamic fictitious play in a potential game and two climbing hill games, one with two players and the other with three players. Our algorithm converges more quickly to the optimum thanboth the competitor algorithms. Hence by placing a greater computational demand on the individual agents, less communication is required between the agents.

M3 - Conference contribution/Paper

SN - 9780981738109

BT - Optimization in Multi-Agent Systems workshop OptMas, 17th International Conference on Autonomous Agents and Multiagent Systems AAMAS 2008.

A2 - Padgham, Lin

A2 - Parkes, David

A2 - Müller, Jörg

PB - INESC-ID

ER -