Differential Evolution Algorithm with Incremental Social Learning
Serdar Özyön · Bilecik Şeyh Edebali Üniversitesi Fen Bilimleri Dergisi · 2020
In this study, the differential evolution algorithm (DE), which has a very strong place among the optimization algorithms in literature, has been tried to be improved and bettered. The algorithm has been bettered by integrating incremental social learning (ISL) structure, which was applied to different optimization algorithms previously with positive feedbacks, into DE. In this betterment, DE has been initiated to search with a number of determined individuals, new individuals have been added to the population with different approaches in certain levels, the process of adding individuals has been ended at the maximum population number determined and the search has been continued with this population number until the stopping criterion has been provided. This new bettered algorithm which has been revealed as a new version of DE has been called Incremental differential evolution algorithm (IDE). Another purpose that comes into prominence in the study is to determine the best method to add individuals in ISL structure. For this purpose, five different approaches have been used in the operation of adding individuals to DE. A set of 13 unimodal and multimodal test functions defined on a 30-dimensional space have been solved with DE and IDE algorithms improved in this study. Evaluations have been made by examining the obtained numerical results, graphics and statistical analyses.