THE AGGREGATION MECHANISM FOR AGENT-BASED EVOLUTIONARY COMPUTATION

Evolutionary algorithms have some indispensable features, which limits their applicability in the case of selected problems. The key concept of the agent-based evolutionary computation paradigm is the decentralization of the evolutionary processes. Such decentralized models of evolutionary computation have some interesting features, which are absent in “classical” evolutionary algorithms— among them there are the possibilities of introducing in a very natural and coherent way new mechanisms and operators. One of such operators—the aggregation—is presented in this paper. Also, there are presented results of preliminary experiments with the evolutionary multi-agent system with aggregation mechanism for time series prediction.