From Adaptive to More Dynamic Control in Evolutionary Algorithms精品.pdfVIP

From Adaptive to More Dynamic Control in Evolutionary Algorithms 1 1 2 1 Giacomo di Tollo , Fr´ed´eric Lardeux , Jorge Maturana , and Fr´ed´eric Saubion 1 LERIA, University of Angers, France name@info.univ-angers.fr 2 Instituto de Inform´atica, Universidad Austral de Chile, Chile jorge.maturana@inf.uach.cl Abstract. Adaptive evolutionary algorithms have been widely devel- oped to improve the management of the balance between intensification and diversification during the search. Nevertheless, this balance may need to be dynamically adjusted over time. Based on previous works on adap- tive operator selection, we investigate in this paper how an adaptive controller can be used to achieve more dynamic search scenarios and what is the real impact of possible combinations of control components. This study may be helpful for the development of more autonomous and efficient evolutionary algorithms. 1 Introduction From a high level point of view, Evolutionary Algorithms (EAs) [5] manage a set (population) of possible configurations of the problem solutions(individuals), which are progressively modified by variation operators, in order to converge to an optimal solution or, at least, to a sub-optimum of good quality. EAs have been successfully applied to various combinatorial and continuous optimization problems. It is now clearly assessed that the successful use of an EA mainly relies on the suitable combination of several components: the choice of an ap- propriate encoding of the problem, the definition of efficient operators and the adjustment of the behaviour of the algor