Session:

LBP - Late Breaking Papers

Title:

Towards a Generally Applicable Self-adapting Hybridization of Evolutionary Algorithms

Authors:

Wilfried Jakob
Christian Blume
Georg Bretthauer

Abstract:

When applied to real-world problems, the powerful optimization tool of Evolutionary Algorithms frequently turns out to be too time-consuming due to elaborate fitness calculations that are often based on run-time-intensive simulations. Incorporating domain-specific knowledge by problem-tailored heuris-tics or local searchers is a commonly used solution, but turns the generally applicable Evolutionary Algorithm into a problem-specific tool. The new method of hybridization implemented in HyGLEAM is aimed at overcoming this limitation and getting the best of both algorithm classes: A fast, globally searching, and robust procedure that preserves the convergence reliability of evolutionary search. Extensive tests demonstrate the superiority of the approach, but also show a drawback: No common parameterization can be drawn from the experiments. As a solution, a new concept of a self-adapting hybrid is introduced. It is stressed that the methods presented can be applied to Evolutionary Algorithms other than the one used here with no or minor modifications being required only.