New Insights into Fuzzy Genetic Algorithms for Optimization Problems

Abstract

In this paper, we shed light on the use of two types of fuzzy genetic algorithms, which stand out from the literature due to the innovative ideas behind them. One is the Gendered Fuzzy Genetic Algorithm, where the crossover mechanism is regulated by the gender and the age of the population to generate offspring through proper fuzzy rules. The other one is the Elegant Fuzzy Genetic Algorithm, where the priority of the parent genome is updated based on the child’s fitness. Both algorithms present a significant computational burden. To speed up the computation, we propose to adopt a nearest-neighbor caching strategy. We first performed several experiments, using some well-known benchmark functions, and tried different types of membership functions and logical connectives. Afterward, some additional benchmarks were retrieved from the literature for a fair comparison against published results, which were obtained by means of former variants of fuzzy genetic algorithms. A real-world application problem, which was retrieved from the literature and dealt with rice production, was also tackled. All the numerical results show the potential of the proposed strategy.