Performance Analysis of Nature Inspired Computing Algorithms Under Hard Restrictions

Nature inspired computing (NIC) provides solutions to complex problems with imitating the behavior of nature or living creature in nature. NIC gives a point of new perspective to various search areas such as; evolutionary algorithms, swarm intelligence, neural network etc. Particle Swarm Optimization (PSO), Artificial Bee Colony (ABC) and Grey Wolf Optimizer (GWO) are among the well-known swarm-based optimization algorithms. These algorithms have demonstrated substantially successful outcomes and have been the subject of various studies. The main motivation of this study is, however, to examine their performances under severely hard restrictions. Restrictions are applied by limiting the number of iterations and the swarm size to very small numbers to observe the algorithms' convergence capabilities. Three scenarios for SwarmSize/NumberOfIterations; 5/24 10/12 and 15/8 are examined. In all scenarios, a total number of evaluations are kept same to reliably compare the results. The algorithms are tested on five unimodal and five multimodal benchmark test functions. The test results revealed that ABC is the best performing algorithm in all scenarios comparing to PSO and GWO.