SHADE algorithm dynamic analyzed through complex network

In this preliminary study, the dynamic of continuous optimization algorithm Success-History based Adaptive Differential Evolution (SHADE) is translated into a Complex Network (CN) and the basic network feature, node degree centrality, is analyzed in order to provide helpful insight into the inner workings of this state-of-the-art Differential Evolution (DE) variant. The analysis is aimed at the correlation between objective function value of an individual and its participation in production of better offspring for the future generation. In order to test the robustness of this method, it is evaluated on the CEC2015 benchmark in 10 and 30 dimensions. © 2017, Springer International Publishing AG