Computational Modeling Lab

Membership Details (University/College)

Contact: Prof. Ann Nowe
Member node: 059
Vrije Universiteit Brussel WE - DINF - COMO Pleinlaan 2 1050 Brussel

>>> Include yourself at this organisation
>>> Update your name or personal details listed below
>>> Submit papers, projects etc to the archive
>>> Update Member Description or Contact details


COMO stands for computational modeling: This is on the one hand the modeling of natural phenomena, and on the other hand developing algorithms for complex problem solving inspired by these natural phenomena. COMO has experience in a wide range of learning techniques such as Reinforcement Learning, Genetic Algorithms, Neural Networks, Support Vector Machines, Bayesian Networks, etc. Nowadays, research in COMO is organized around two research tracks: 1) machine learning techniques for data mining applications, and 2) multi-agent systems (MAS).

Concerning the second track, we strongly focus an evolutionary view on MAS and on the problem of learning and adaptation in MAS.

In the evolutionary approach, we use biological metaphors like natural selection, co-evolution and evolutionary transitions to investigate problems like how collaborative or cooperative behavior can emerge in a complex environment of interacting agents that compete for the limited available resources. In evolutionary biology, we often see cooperation at the group level although the individuals are competing for limited resources. This dilemma has been overcome several times in evolutionary transitions like for instance from unicellular organisms to multi-cellular ones. Our goal here is to find the necessary and sufficient conditions under which cooperation can emerge between competing individuals and to use it in the context of MAS.

The problems of learning and adaptation in MAS are studied from the perspective of existing machine learning techniques especially reinforcement learning (RL). RL is well-understood for stationary environments, both theoretically and empirically. In contrast, in MAS we face state dependent non-stationary environments, i.e. environments that are non-stationary not only because conditions are changing in the environment, but also because actions taken by other agents influence the rewards an agent experience for taken some action. Our goal here is to get more insight in the effect of state dependent non-stationarity on the application of RL techniques and how RL can be adapted for learning in MAS.

Both above directions are joined in a theoretical framework based on evolutionary game theory (EGT) and which we are currently developing. EGT was developed in evolutionary biology to understand among other things phenomena like evolutionary transitions.

The experience of COMO in the application of machine learning techniques to telecom problems such as distributed load-based routing will certainly be of great value for the proposed project. Routing is an interesting case study for learning in MAS, it possess the important properties of a MAS because it is distributed, communication is delayed and not for free, routers have to take autonomously decisions based on limited information, and the state dependent non-stationarity is very apparent. These properties are also present in a learning MAS for traffic management.

The experience of COMO in evolutionary computation, evolutionary game theory and the theory of complex dynamical systems will contribute to the understanding of the dynamics and the evolution of MAS including their emergent properties.

For a list of publications, please visit our website at

Book Section

Gonzalez, P. and Negrete, J. and Barreiro, A. and Gershenson, C. (2000) A Model for Combination of External and Internal Stimuli in the Action Selection of an Autonomous Agent. In: MICAI 2000: Advances in Artificial Intelligence. Lecture Notes in Artificial Intelligence, 1793. Springer-Verlag, pp. 621-633.

Gershenson, C. and Gonzalez, P. and Negrete, J. (2000) Action Selection Properties in a Software Simulated Agent. In: MICAI 2000: Advances in Artificial Intelligence. Lecture Notes in Artificial Intelligence, 1793. Springer-Verlag, pp. 634-648.

Gershenson, C. and Gonzalez, P. and Negrete, J. (2000) Thinking Adaptive: Towards a Behaviours Virtual Laboratory. In: SAB2000 Proceedings Supplement Book. International Society for Adaptive Behavior, Honolulu.


[Course] NOWE, Ann (2006) Multiagent systems.


Gershenson, C. (2001) Artificial Societies of Intelligent Agents. Other thesis, Fundación Arturo Rosenblueth.

This list was generated on Mon Sep 10 16:38:56 BST 2007.


AgentLink is the European Commission's IST-funded Coordination Action for Agent-Based Computing
and is coordinated by the
University of Liverpool and University of Southampton
If you encounter any problems with these pages please contact