Cooperative model-based metaheuristics

The paper presents a methodology for construction of cooperative model-based metaheuristics for combinatorial optimization problems. Its distinctive feature is two-level structure. The lower level consists of independent model-based algorithms called basic. The higher (guiding) level performs search in the a space of problem models using individual models provided by basic algorithms. The guiding procedure also implements search experience exchange between basic algorithms. As basic algorithms one can take copies of the algorithms that belong either to the same or to the different model-based methods. Cooperative metaheuristic based on ant colony optimization was developed for study of the approach. Results of computational experiment show the effectiveness of the suggested cooperation scheme. Presented framework can serve as a basis for hyper-heuristics development.