User-Assisted Management of Agent-Based Evolutionary Computation

In the paper the need and general idea of user-assisted management of computational intelligence systems is discussed. First some methodological issues are presented with particular attention paid to agent-based approaches. These general considerations are supported by the case of evolutionary multi-agent system (EMAS) with immunological selection, applied to multi-modal function optimization subjected to a user-driven tuning procedure. Finally, preliminary experimental results considering the influence of selected parameters on the performance of the system are shown.