C. Breitschopf, G. Blaschek, T. Scheidl: Adaptive Guidance of the Search Process in Evolutionary Optimization, Proceedings of the 4th WSEAS Int. Conference on Computational Intelligence, Man-Machine Systems and Cybernetics (CIMMACS '05), ISSN: 960-8457-38-6, Miami, Florida, USA, November 17-19, 2005.
Evolutionary optimization is a well-known paradigm for solving large-scale combinatorial optimization problems. Evolutionary algorithms typically consider the fitness of solutions to decide which solution should be processed by an operator. In the presence of multiple operators to choose from, similar strategies are needed to choose an appropriate operator. In this paper, we present an adaptive target-oriented approach for evaluating and selecting operators on the fly. This technique has been integrated into the OptLets framework, which monitors the success of operators and uses the results of this evaluation for operator selection in the future. Although this paper describes the technique and illustrates the results in the context of the OptLets framework, the evaluation strategy is applicable for other population-based optimization systems as well.