METAHEURISTIC ALGORITHMS. METAPHORS-STRATEGIES (REVIEW ARTICLE)

Abstract

The relevance of modern metaheuristic algorithms, clarification of a number of terms and relationships between them, the need for classification, as well as metaphors used to describe algorithms are described. This makes it possible to understand the need to cover the specified topic and conduct research on literary sources related to the issue. A number of terms and categories and their interrelationships were considered, which made it possible to propose a classification of metaheuristic algorithms. An approach to the classification of metaheuristic algorithms based on the terms and division of categories taken from the natural sciences is proposed. According to the names of the classes, their filling takes place. This makes it possible to combine a certain segment of knowledge into a cluster with a single terminology. Considering the category of "metaphor" and its functions in the formation of metaheuristic algorithms, it made it possible to gain a deeper understanding of the possibilities of using metaphors in scientific activity and to form a list of requirements for them when describing algorithms. An overview of interesting, original and diverse metaheuristic algorithms was conducted, which made it possible to understand modern trends in this issue, to determine the advantages and disadvantages of algorithms, as well as to understand and shape the role of metaphor in their formation or description. Also, the review makes it possible to distinguish two intellectual directions of using metaphors: helping to increase understanding and intensifying the delivery of the idea to the target audience of an already developed algorithm or strategy and the development of new algorithms or strategies for finding optimal parameters. The system of formation and design of new knowledge and the role of metaphor in it are considered. The system triangle "idea-algorithm-metaphor" has been formed, as well as possible ways of development in this system, which allows the developer to define or choose a certain path and its subsequent stages. Keywords: metaheuristic algorithm, metaphor, optimization.