Visualization of optimal solutions using self-organizing maps in computational electromagnetism

In this paper, the self-organizing map is applied for visualization and clustering of optimal solutions in parameter spaces. The present method is tested for the simplified version of De Jong´s function and TEAM workshop problem 22. It is shown that the present method successfully visualizes the distribution of optimal solutions, and allows us to classify the optimal solutions into clusters. Moreover, the optimal regions which include clustered optimal solutions in the parameter space can be obtained by the present method.