The paper deals with the method for searching the proper values of behavioural (relevant) parameters of optimization algorithms for large-scale problems. The authors formulate the optimization task as multi-objective problem taking into account two criteria. The first criterion corresponds to the estimation of the accuracy of a solution, whereas the second one represents the time computational complexity of the main optimization algorithm. In the present study, predominant Pareto optimality concept is used to solve this problem. Moreover, the authors propose to use a much less complicated algorithm in the main optimization engine, while a more advanced approach in the meta-evolution core. The engine of the target optimization algorithm is realised applying the particle swarm optimization algorithm, while the core of the meta-evolution process is implemented by means of the multi-objective evolutionary algorithm. The advantages and limitations of the proposed meta-evolution method were examined employing well-practised test functions described in the literature.
CITATION STYLE
Przystałka, P., & Katunin, A. (2016). Multi-objective meta-evolution method for large-scale optimization problems. Studies in Computational Intelligence, 610, 165–182. https://doi.org/10.1007/978-3-319-21133-6_10
Mendeley helps you to discover research relevant for your work.