Entropy driven self-adaptive Differential Evolution

Citation data:

MENDEL 2008 - 14th International Conference on Soft Computing: Evolutionary Computation, Genetic Programming, Fuzzy Logic, Rough Sets, Neural Networks, Fractals, Bayesian Methods, Page: 38-43

Publication Year:
2008
Usage 270
Abstract Views 267
Downloads 3
Repository URL:
http://publikace.k.utb.cz/handle/10563/1001871; http://hdl.handle.net/10563/1001871
Author(s):
Běhal, Ladislav; Vlček, Karel
Publisher(s):
Vysoké učení technické v Brně
Tags:
Computer Science; entropy; Gaussian entropy; linear entropy; differential evolution; self-adapting
conference paper description
Self-adaptiveDE is one of the most efficient and popular modifications of evolution algorithm (Differential Evolution). The study deals with the possibility of using the entropy evaluation of population for SaDE algorithm modification, with an impact on its convergence towards global extreme. Further, the unavoidable increase of computational complexity is discussed. The paper proposes an algorithm based on the efficiency principles of SaDE algorithm. This proposed algorithm uses the above mentioned entropy calculation. In order to increase the algorithm efficiency, a recursive differential evolution was applied; it modifies its scale factors to gain the best convergence abilities, with respect to its robustness. Detailed performance comparisons of this algorithm on the currently used benchmark functions are included.