Impact of Structural Bias on the Sine Cosine Algorithm: A Theoretical Investigation Using the Signature Test

Metaheuristic algorithms have been recognized for their effectiveness in solving non-convex and non-linear complex optimization problems. These algorithms are influenced by landscape bias, guided by objective function values, and algorithmic operator bias directed by the operator used in the algorithms. The presence of algorithmic operator bias, also known as structural bias, forces the population to revisit a particular region, badly affecting the algorithm’s exploration capacity. Also, since the population revisits the same place without gaining new information, it increases computational costs and slows the convergence rate. Therefore, it is crucial to identify and address structural bias to enhance algorithm performance and reduce computational time. To the best of our knowledge, no previous study has focused on investigating the structural bias of the Sine Cosine Algorithm (SCA) in the existing literature. Therefore, the main objective of this study is to examine the structural bias present in the SCA, a widely used metaheuristic algorithm. To investigate structural bias signature test is employed. Additionally, average Euclidean distances of the population is calculated to assess spatial relationships and overall distribution. Our analysis uncovers a prominent bias in the SCA towards the axes and the origin, suggesting a strong tendency to converge towards specific regions within the search space. By understanding and characterizing this bias, we provide valuable insights into the behavior of the SCA, which can contribute to the research community’s understanding and guide future improvements in algorithm design.