Comparative Study of Cuckoo-Inspired Algorithms to Solve Large-Scale Continuous Optimization Problems

Two distinctive behaviors of the cuckoo bird have inspired several metaheuristic algorithms to solve continuous optimizationproblems. In addition to the well-known parasitic breeding behavior that gave rise to several cuckoo search (CS) algorithms, another behavior related to their clustering and the way they locate food sources has given rise to the COA algorithm. As a result, there are several variants to solve continuous optimization problems; however, it is necessary to define which one is the most suitable under specific requirements. This paper compares six of these algorithms, including CS+LEM (proposed in this paper), which consists of a hybridization of the CS algorithm with learning evolutionary models (LEM) using an approach known as “metaheuristics enhanced by artificial intelligence”. Three assessments were performed using a set of 61 continuous test functions: 1) the optimal value achieved with a fixed execution time; 2) the number of objective function evaluations required to reach the global optimum; and 3) the optimal value achieved with a fixed number of objective function evaluations. CS+LEM presents the best results in evaluation 1, while COA presents the best results in evaluations 2 and 3. The results were analyzed using the Friedman and Wilcoxon nonparametric statistical tests.