The Combination of Estimation of Distribution With Differential Evolution (EDDE)
Tarik Eltaeib, Suhil M. M Elsbay, Nazrul Islam · 2024
Estimation distribution of Differential Evolution optimizes the algorithm by monitoring data of the population of potential solutions. The real population does not need to be maintained from one generation to the next since the population statistics are kept up to date. Generating new candidates, the population is based on adding a scaled version of the difference vector between two individuals to the next population. The population is produced from previous generation data each generation, and using various definitions of relation, the information of the population's fittest individuals is calculated. This procedure is continued from one generation to the next. Therefore, the algorithm is population-based and discards a portion of the population every generation, replacing them with individuals with statistical characteristics that are reasonably fit. This is a new genetic algorithm approach that works not on randomly generating new populations. It provides the density of elite populations based on the statistical properties of the high-fit individuals.