About this project
The process of identifying and modelling the appropriate agent behavior is a cumbersome task -- no matter whether it is for agent-based software or agent-based simulation.
In Agent-based Simulation this is worsened by the fact that modeling happens on the agent level, while the intended behavior is produced only during a simulation run.
Machine Learning can help by generating partial or preliminary versions of the agent low-level behavior. Yet, just adapting standard methods is not sufficient, as for actually being useful for a human modeler needs to be enabled to analyse the learnt behavior which should never be just a black box optimizing some more or less well defined objective function. In this project we develop and test learning methods that support the agent modelling task by delegating the design to the adaptive agents.