Ensemble of Different Parameter Adaptation Techniques in Differential Evolution
Differential evolution has been proved to be one of the most powerful evolutionary algorithms for the numerical optimization. However, the performance of differential evolution is significantly influenced by its parameter settings. To remedy this limitation, different parameter adaptation techniques are proposed in the literature. Generally, different parameter adaptation techniques have different rationales and may be suitable to different problems. Based on this consideration, in this paper, we attempt to develop the ensemble of different parameter adaptation techniques to enhance the performance of differential evolution. In our proposed method, different parameter adaptation techniques are combined together to adjust the parameters of different solutions in the population. As an illustration, two parameter adaptation techniques proposed in the literature are used in our proposed method. To verify the performance of our proposal, the functions proposed in CEC 2005 are chosen as the test suite. Experimental results indicate that, on the whole, our proposed method is able to provide better results than the single parameter adaptation based differential evolution variants with respect to the non-parametric statistical tests.
- 4.Gämperle, R., Müler, S., Koumoutsakos, P.: A parameter study for differential evolution. In: Proceedings of the WSEAS International Conference Advances in Intelligent Systems, Fuzzy Systems, Evolutionary Computation, pp. 293–298 (2002)Google Scholar
- 10.Tanabe, R., Fukunaga, A.: Success-history based parameter adaptation for differential evolution. In: 2013 IEEE Congress on Evolutionary Computation (CEC), pp. 71–78 (2013)Google Scholar
- 11.Suganthan, P.N., Hansen, N., Liang, J.J., Deb, K., Chen, Y.P., Auger, A., Tiwari, S.: Problem definitions and evaluation criteria for the CEC_2005 special session on real-parameter optimization (2005)Google Scholar
- 12.Brest, J., Zumer, V., Maucec, M.: Self-adaptive differential evolution algorithm in constrained real-parameter optimization. In: IEEE Congress on Evolutionary Computation, pp. 215–222 (2006)Google Scholar