Fuzzy sets and fuzzy logic: theory and applications
Fuzzy sets and fuzzy logic: theory and applications
Centroid of a type-2 fuzzy set
Information Sciences: an International Journal
Calculating the expected loss of diversity of selection schemes
Evolutionary Computation
Predictive models for the breeder genetic algorithm i. continuous parameter optimization
Evolutionary Computation
A comparison of selection schemes used in evolutionary algorithms
Evolutionary Computation
ICAISC'10 Proceedings of the 10th international conference on Artifical intelligence and soft computing: Part II
IEEE Transactions on Fuzzy Systems
Hi-index | 0.00 |
An optimal trade-off between exploration and exploitation properties of genetic algorithm is very important in optimization process. Due value steering of these two factors we can prevent a premature convergence of the algorithm. Therefore, better results can be obtained during optimization process with the use of genetic algorithm. In this paper the type-2 fuzzy logic control of trade-off between exploration and exploitation properties of genetic algorithm is presented. Our novel selection method (with application of type-2 fuzzy logic to steering of key parameter in this selection method) is based on previously elaborated mix selection method. In proposed method two factors are taken into consideration: the first is a number of generations of genetic algorithm, and second is a population diversity. Due to these two factors, we can control the trade-off between global and local search of solution space; also due to the type-2 fuzzy control the proposed method is more "immune" in falling into the trap of local extremum. The results obtained using proposed method (during optimization of test functions chosen from literature) are compared with the results obtained using other selection methods. Also, a statistically importance of obtained results is checked using statistical t-Student test. In almost all cases, the results obtained using proposed selection method are statistically important and better than the results obtained using other selection techniques.