Population Size Management in a Cuckoo Search Algorithm Solving Combinatorial Problems

As is well known in the scientific community, optimization problems are becoming increasingly common, complex, and difficult to solve. The use of metaheuristics to solve these problems is gaining momentum thanks to their great adaptability. Because of this, there is a need to generate robust metaheuristics with a good balance of exploration and exploitation for different optimization problems. Our proposal seeks to improve the exploration and exploitation balance by incorporating a dynamic variation of the population. For this purpose, we implement the Cuckoo Search metaheuristic in its two versions, with and without dynamic population, to solve 3 classical optimization problems. Preliminary results are very good in terms of performance but indicate that it is not enough to vary the population dynamically, but it is necessary to add additional perturbation operators to force changes in the metaheuristic behavior.