Kühn, E., & Šešum-Čavić, V. (2022). A Framework-Based Approach for Flexible Evaluation of Swarm-Intelligent Algorithms. In A. E. Smith (Ed.), Women in Computational Intelligence (pp. 393–412). Springer. https://doi.org/10.1007/978-3-030-79092-9_18
E194-05 - Forschungsbereich Compilers and Languages
Women in Computational Intelligence
Number of Pages:
methodology for evaluation of algorithms; recommendation of algorithm for given use case; coordination middleware; Self-organization; Distributed middleware architectures; Coordination Patterns
Many important Information Technology (IT) problems comprise some kind of optimization task that is challenging from the computation point of view and therefore needs to be addressed by an appropriate metaheuristic algorithm. Recently, a diversity of promising swarm-inspired metaheuristics appeared. However, the decision about what algorithm fits best for which class of problems and the fair comparison of a variety of algorithms are not clearly articulated from the methodological point of view. The usual approach taken is to integrate a certain algorithm into the solution for a concrete problem and to evaluate its usefulness. The deficiency of this approach is that such specialized solutions hardly allow the comparison of different algorithms under changing environmental conditions. In this chapter, we propose a new methodology for the evaluation of algorithms that are suited for highly dynamic scenarios requiring self-organization. The solution is based on a coordination middleware architecture that is used for the implementation of flexible and pattern-based frameworks and test beds. The evaluation is carried out by means of several use cases.