Home > Research > Publications & Outputs > A game-theoretic calibration approach for agent...

Links

Text available via DOI:

View graph of relations

A game-theoretic calibration approach for agent-based planning simulations

Research output: Contribution to journalJournal articlepeer-review

Published

Standard

A game-theoretic calibration approach for agent-based planning simulations. / Buwaya, Julia; Cleophas, Catherine.

In: IFAC-PapersOnLine, Vol. 48, No. 1, 01.02.2015, p. 844-849.

Research output: Contribution to journalJournal articlepeer-review

Harvard

APA

Vancouver

Author

Buwaya, Julia ; Cleophas, Catherine. / A game-theoretic calibration approach for agent-based planning simulations. In: IFAC-PapersOnLine. 2015 ; Vol. 48, No. 1. pp. 844-849.

Bibtex

@article{9bb6fe837f954e58aba61f5ed349d06a,
title = "A game-theoretic calibration approach for agent-based planning simulations",
abstract = "Simulations are increasingly employed to evaluate alternative planning strategies. With the number of model parameters and their interdependencies grow challenges regarding the validation and calibration of simulation systems. A major challenge lies in calibrating models that include emergent phenomena, as is a frequently stated feature of agent-based simulations. Here heterogeneous groups of agents are directly modeled to enable the consideration of agents' impact on the planning solution and its success. Difficulties in calibration arise as the system's behavior emerges from agents' individual decisions and actions, which cannot be fully observed. In the following we present a novel approach for calibrating agent input parameter values of agent-based simulations for decision support. The approach is based on a game-theoretic model representing an approximation of the dynamic simulation system. Its performance is tested in a meta-simulation framework on the example of a market simulation with a monopolist supplier. The meta-simulation setting allows us to compare calibration results to actual input parameters. The presented work provides an outline for efficient calibration of agent-based simulations in general using a game-theoretic model.",
keywords = "Agents, Calibration, Decision support systems, Game theory, Integer programming, Simulation",
author = "Julia Buwaya and Catherine Cleophas",
year = "2015",
month = feb,
day = "1",
doi = "10.1016/j.ifacol.2015.05.044",
language = "English",
volume = "48",
pages = "844--849",
journal = "IFAC-PapersOnLine",
issn = "2405-8963",
publisher = "IFAC Secretariat",
number = "1",

}

RIS

TY - JOUR

T1 - A game-theoretic calibration approach for agent-based planning simulations

AU - Buwaya, Julia

AU - Cleophas, Catherine

PY - 2015/2/1

Y1 - 2015/2/1

N2 - Simulations are increasingly employed to evaluate alternative planning strategies. With the number of model parameters and their interdependencies grow challenges regarding the validation and calibration of simulation systems. A major challenge lies in calibrating models that include emergent phenomena, as is a frequently stated feature of agent-based simulations. Here heterogeneous groups of agents are directly modeled to enable the consideration of agents' impact on the planning solution and its success. Difficulties in calibration arise as the system's behavior emerges from agents' individual decisions and actions, which cannot be fully observed. In the following we present a novel approach for calibrating agent input parameter values of agent-based simulations for decision support. The approach is based on a game-theoretic model representing an approximation of the dynamic simulation system. Its performance is tested in a meta-simulation framework on the example of a market simulation with a monopolist supplier. The meta-simulation setting allows us to compare calibration results to actual input parameters. The presented work provides an outline for efficient calibration of agent-based simulations in general using a game-theoretic model.

AB - Simulations are increasingly employed to evaluate alternative planning strategies. With the number of model parameters and their interdependencies grow challenges regarding the validation and calibration of simulation systems. A major challenge lies in calibrating models that include emergent phenomena, as is a frequently stated feature of agent-based simulations. Here heterogeneous groups of agents are directly modeled to enable the consideration of agents' impact on the planning solution and its success. Difficulties in calibration arise as the system's behavior emerges from agents' individual decisions and actions, which cannot be fully observed. In the following we present a novel approach for calibrating agent input parameter values of agent-based simulations for decision support. The approach is based on a game-theoretic model representing an approximation of the dynamic simulation system. Its performance is tested in a meta-simulation framework on the example of a market simulation with a monopolist supplier. The meta-simulation setting allows us to compare calibration results to actual input parameters. The presented work provides an outline for efficient calibration of agent-based simulations in general using a game-theoretic model.

KW - Agents

KW - Calibration

KW - Decision support systems

KW - Game theory

KW - Integer programming

KW - Simulation

U2 - 10.1016/j.ifacol.2015.05.044

DO - 10.1016/j.ifacol.2015.05.044

M3 - Journal article

VL - 48

SP - 844

EP - 849

JO - IFAC-PapersOnLine

JF - IFAC-PapersOnLine

SN - 2405-8963

IS - 1

ER -