СОВРЕМЕННОЕ СОСТОЯНИЕ БИОЭВРИСТИК: КЛАССИФИКАЦИЯ, БЕНЧМАРКИНГ, ОБЛАСТИ ПРИМЕНЕНИЯ

Abstract

The purpose of this article is to analyze the current state of research in the field of developmentof algorithms inspired by nature, including categorization, classification, testing, citation,and application areas. A new multi-level classification system based on the following features ispresented: the criterion of conformity to a natural metaphor, structural, behavioral, search, component,and evaluation criteria. The classification of bio heuristics involves the systematic assignmentof each bio heuristics to one and only one class within a system of mutually exclusive andnon-overlapping classes. Categorization allows an objective approach to the choice of bio heuristics.For each bio heuristics there are specific tasks with which it copes well. Knowing these relationshipsis important for the purposeful application of bio heuristics. An example of classificationis considered. It is noted that the most informative classification criterion is the behavioral criterion,the most cited class of bio heuristics are swarm intelligence algorithms, and the most cited bioheuristics is the PSO particle swarm algorithm. Modern approaches to benchmarking of bio heuristicsare presented: discrete and continuous optimization problems, as well as optimization engineeringproblems. There is a tendency to compare the performance of bio heuristics using statisticalhypothesis testing on benchmarks. The tasks successfully solved by bio heuristics in such areasas engineering design, image processing and computer vision, computer networks and communications,energy and energy management, data analysis and machine learning, robotics, medicaldiagnostics are systematized. There is a tendency to hybridize bio heuristics in one optimizer.However, convincing evidence is required that the results compensate for the increase in complexitycompared to individual algorithms. Optimization problems requiring further research are noted:dynamic and stochastic optimization problems; multicriteria optimization problems; multimodaloptimization problems; multidimensional optimization problems; memetic optimizationproblems in which a variety of search algorithms are combined; optimization problems and adaptationof bio heuristics parameter settings to achieve a balance between the convergence rate andthe diversification of the solution search space.