A Multi-population-Based Algorithm with Different Ways of Subpopulations Cooperation

A New Approach to Detecting and Preventing Populations Stagnation Through Dynamic Changes in Multi-Population-Based Algorithms

“…[8]) or topologies where the best or worst population is placed in a specific location (see e.g. [12]). Moreover, different approaches, such as a fitness-based Migration Policy designed to promote the maintenance of diversity through a mechanism that combines groups of individuals to alternate between exploration and exploitation proposed in [6], can be used instead.…”

“…An interesting overview and performance tests of 36 different migration topologies is presented in [12]. The best topologies, regardless of other configurations and simulation problems used, turned out to be: an inverted star with a mechanism for setting the best population in the middle of topology (see Figure 1.a)), a ladder with a down-top connection (see Figure 1.b)), and a topology in which each population has two random connections with others (see Figure 1.c)).…”

“…The paper [12] shows the influence of migration parameters on a given migration topology. The best parameters turned out to be the frequency of replacement every 10 iterations of the algorithm, and the selection of an individual for replacement using the roulette wheel method.…”

“…The paper [12] shows that, regardless of the migration topology and the simulation problem under consideration used, a good configuration is a topology with 8 islands composed of 32 individuals each, while the paper [40] shows that algorithms in which islands operate based on different population algorithms allow obtaining better and more stable results.…”