Complex Network Analysis in PSO as an Fitness Landscape Classifier
Department or Administrative Unit
In this paper, an initial small-scale study is carried out. It is proposed that using the complex network analysis it may be possible to make a classification of the fitness landscape type. A complex network is constructed from the inner dynamics of the population in PSO algorithm. The mean and maximal number of links in the network is then evaluated alongside with other basic statistic characteristics. It is shown on a basic function set that the number of links in the networks may vary significantly when facing unimodal and multimodal problems. Initial visualizations of the constructed complex networks are presented and the results are discussed with proposals for future research and possible future applications of this method.
Pluhacek, M., Senkerik, R., Janostik, A. V. J., & Davendra, D. (2016). Complex network analysis in PSO as an fitness landscape classifier. 2016 IEEE Congress on Evolutionary Computation (CEC). https://doi.org/10.1109/cec.2016.7744211
2016 IEEE Congress on Evolutionary Computation (CEC)
Copyright © 2016, IEEE
This article was originally published in 2016 IEEE Congress on Evolutionary Computation (CEC). The full-text article from the publisher can be found here.
Due to copyright restrictions, this article is not available for free download from ScholarWorks @ CWU.