Empirical Evidences to Validate the Performance of Self-Switching Base Vector Based Mutation of Differential Evolution Algorithm

摘要

There exist many tools in Computer Science to solve the optimization problems around us. One such tool is the set of algorithms known as Evolutionary Algorithms (EAs) which is under the Evolutionary Computing (EC) field. The Differential Evolution (DE) algorithm in the set of EAs is known for its unique mutation scheme. There are many research works in the literature to further study and modify this scheme to propose new mutation schemes. This paper presents detailed and extensive empirical evidences for the Self-Switching Base Vector Selection base mutation scheme (termed as DE/randorbest/1) found in the literature. The results for our validation are obtained by running the DE algorithm for all possible values of its control parameters (Mutation Step Size (F) and Crossover Rate (C_r)) on a Benchmarking-Function suite. The obtained results are compared by the performance metrics: Average Solution Accuracy (ASA) and Success Rate (SR).