volker

Personal

Full Name

Volker Nannen

Homepage

http://volker.nannen.com

Position

Ph.D. student in Computer Science and Economics

Affiliation

Vrije Universiteit Amsterdam
ISI Foundation, Turin

Research topic

Evolution of strategies in agent based economic simulations.

Research interests

Multi-agent systems, information theory, complex systems.

Ph.D. project

Thesis title

Evolutionary Agent-Based Policy Analysis in Dynamic Environments

Defense year

2009

Faculty

Faculty of Computer Science, Vrije Universiteit Amsterdam

Ph.D. Supervisors

Prof. Dr. J.C.J.M van den Bergh
Faculty of Economics and Business Administration &
Institute for Environmental Studies, Vrije Universiteit Amsterdam

Prof. Dr. A.E. Eiben
Faculty of Computer Science, Vrije Universiteit Amsterdam

General project description

This is an interdisciplinary research project between the departments of Economics and Computer Science at the Vrije Universiteit of Amsterdam. The official title of the research project is 'Impact of environmental dynamics on social systems with behavioral interactions: evolutionary computational experiments.'

Research question

A major obstacle in building an evolutionary agent-based model for economics is to find a plausible model of social interaction. Besides reflecting obvious observations on how real agents exchange information, such a model also has to be computationally robust and consistent.

Motivation

Standard economic models are based on a representative homogeneous agent and use aggregate production functions. That makes it impossible to design and evaluate policies that differentiate between agents and selectively encourage or discourage certain behavior. In an evolutionary agent-based simulation we can do exactly that. We evalute environmental policies that target specific agents and study how this effects the adoption rate of their behavior.

Approach and methods

We work on new statistical and numerical methods to study the robustness of evolutionary models.

Contributions

We have developed REVAC, a numeric method to measure parameter relevance in terms of the information needed to calibrate a parameter. This has enabled us to build a robust model of social evolution. We have used this stable model to design and numerically evaluate new environmental policies, the first study of its kind.

Publications

Publications

Volker Nannen, S. K. Smit, and A. E. Eiben, Costs and Benefits of Tuning Parameters of Evolutionary Algorithms, Proceedings of the 20th Conference on Parallel Problem Solving from Nature (PPSN), pp. 528—538, 2008. [pdf]

Volker Nannen and A. E. Eiben, Efficient Relevance Estimation and Value Calibration of Evolutionary Algorithm Parameters, Proceedings of the IEEE Congress on Evolutionary Computation (CEC), pp. 103—110, 2007. [pdf]

W. A. de Landgraaf, A. E. Eiben and V. Nannen, Parameter Calibration using Meta-Algorithms, Proceedings of the IEEE Congress on Evolutionary Computation (CEC), pp. 71—78, 2007. [pdf]

Volker Nannen and A.E. Eiben, Relevance Estimation and Value Calibration of Evolutionary Algorithm Parameters, Proceedings of the Joint International Conference for Artificial Intelligence (IJCAI), pp. 975—980, 2007. [pdf]

Volker Nannen and A.E. Eiben, Generality of Results obtained by the Parameter Calibration and Relevance Estimation Method, Proceedings of the Workshop on Empirical Methods for the Analysis of Algorithms (EMAA), in conjunction with the International Conference on Parallel Problem Solving From Nature (PPSN), Reykjavik, 2006.

Volker Nannen and A.E. Eiben A Method for Parameter Calibration and Relevance Estimation in Evolutionary Processes, Proceedings of the Genetic and Evolutionary Computation Conference (GECCO), pp. 183—190, 2006. [pdf]