Control parameter adaptation strategies for mutation and crossover rates of differential evolution algorithm - An insight

Differential Evolution (DE), an optimization algorithm under the roof of Evolutionary Algorithms (EAs), is well known for its efficiency in solving optimization problems which are non-linear and non-differentiable. DE has many good qualities such as algorithmic simplicity, robustness and reliability. DE also has the quality of solving the given problem with few control parameters (NP - population size, F - mutation rate and Cr - crossover rate). However, suitable setting of values to these parameters is a complicated task. Hence, various adaptation strategies to tune these parameters during the run of DE algorithm are proposed in the literature. Choosing the right adaptation strategy itself is another difficult task, which need in-depth understanding of existing adaptation strategies. The aim of this paper is to summarize various adaptation strategies proposed in DE literature for adapting F and Cr. The adaptation strategies are categorized based on the logic used by the authors for adaptation, and brief insights about each of the categories along with the corresponding adaptation strategies are presented.