Research Information System
15th International Conference on Soft Computing, Brno, Czech Republic, 24 - 26 June 2009, pp.16-22
Hyper-heuristics introduce novel approaches for solving hard combinatorial optimization problems. A hyper-heuristic method operates over a set of low level heuristics. There are different hyper-heuristic frameworks that employ the idea of automating the heuristic design process. In a perturbative hyper-heuristic framework, the most appropriate low level heuristic is automatically determined and applied to solve a given problem at each step of the search process. A landscape analysis technique provides means for understanding the influence of operators and algorithmic behavior for a given problem. In this study, we aim to understand and analyze a set of perturbative hyper-heuristics through landscape analysis based on an auto-correlation function. Tests are performed on a series of commonly used benchmark functions. To the best of the authors' knowledge, no such prior landscape analysis exists in literature for the perturbative hyper-heuristics.