A cooperative system for metaheuristic algorithms. (1st March 2021)

• Record Type: Journal Article
• Authors: Tezel, Baris Tekin; Mert, Ali
• Is Part Of: Expert systems with applications. Volume 165(2021)
• Publication Date: 2021-03-01
• Subjects: Soft computing -- Optimization -- Metaheuristic -- Intelligent system -- Fuzzy inference system; Expert systems (Computer science) -- Periodicals; Systèmes experts (Informatique) -- Périodiques; Electronic journals; 006.33
• DOI: 10.1016/j.eswa.2020.113976
• Languages: English
• ISSNs: 0957-4174

Abstract

Abstract: Optimization problems are defined as the functions whereby the target is to find the optimum state depending on the parameters that have certain limitations. In the field of optimization, the aim is to find from among multiple alternative solutions the optimal solution or approximate solution that provides all the restrictions. Metaheuristic is an extremely effective method to find approximate solutions to optimization problems. However, when metaheuristics are used, there occurs an algorithm selection problem. This problem involves decision-making about which algorithm is to be used to solve the existing optimization problem with maximum performance. The objective of this study is to use a cooperative system that combines different metaheuristics to successfully deal with algorithm selection problems. An intelligent combination of different metaheuristics is expected to provide more flexible, more efficient and more robust approaches. However, such a combination requires less precision. The combination is generated through a methodology designed with soft computing. In addition to the algorithm selection problem, the adjustment of algorithm parameters has significant importance in obtaining good results. For this reason, the cooperative system proposed in this study offers fine-tuning of parameters based on soft computing techniques. Highlights: Creating a cooperative scheme to automatically select a meta-heuristic algorithm. Changing algorithms that cannot promise a better solution and tuning parameters. Controlling the searching process dynamically via fuzzy logic. Using the same amount of computational resources as isolated meta-heuristic.