Enhanced Coati Optimization Algorithm for Big Data Optimization Problem

Abstract

The recently proposed Coati Optimization Algorithm (COA) is one of the swarm-based intelligence algorithms. In this study, the current COA algorithm is developed and Enhanced COA (ECOA) is proposed. There is an imbalance between the exploitation and exploration capabilities of the COA. To balance the exploration and exploitation capabilities of COA in the search space, the algorithm has been improved with two modifications. These modifications are those that preserve population diversity for a longer period of time during local and global searches. Thus, some of the drawbacks of COA in search strategies are eliminated. The achievements of COA and ECOA were tested in four different test groups. COA and ECOA were first compared on twenty-three classic CEC functions in three different dimensions (10, 20, and 30). Later, ECOA was tested on CEC-2017 with twenty-nine functions and on CEC-2020 with ten functions, and its success was demonstrated in different dimensions (5, 10, and 30). Finally, ECOA has been shown to be successful in different cycles (300, 500, and 1000) on Big Data Optimization Problems (BOP), which have high dimensions. Friedman and Wilcoxon tests were performed on the results, and the obtained results were analyzed in detail. According to the results, ECOA outperformed COA in all comparisons performed. In order to prove the success of ECOA, seven newly proposed algorithms (EMA, FHO, SHO, HBA, SMA, SOA, and JAYA) were selected from the literature in the last few years and compared with ECOA and COA. In the classical test functions, ECOA achieved the best results, surpassing all other algorithms when compared. It achieved the second-best results in CEC-2020 test functions and entered the top four in CEC-2017 and BOP test functions. According to the results, ECOA can be used as an alternative algorithm for solving small, medium, and large-scale continuous optimization problems.