Sardine Feast Metaheuristic Optimization with Random Walk for Global Optimization

Abstract

The magnificent sardine feast phenomenon at sea served as the inspiration for the Sardine Feast Metaheuristic Optimization (SFMO) algorithm, which is based on the prevalence of sea predators feeding on sardines. The initial work on the algorithm has shown that it is superior to other optimization algorithms such as Cuckoo Search (CS), Genetic Algorithm (GA), and Bat-inspired Algorithm (BA) in finding global optimization values in benchmark functions. However, SFMO might experience mature convergence because, during exploration and exploitation, it depends on the normal random function in its predators' movement calculation. This paper explores an improvement of SFMO by embedding other random walk algorithms such as Brownian motion and Levy flight. Both random walk algorithms will replace the normal random function to explore divergent areas in the problem space. The performance of the improved SFMO is studied by testing it in several predefined global benchmark functions. The outcomes of the tests are then contrasted with those produced by the basic SFMO algorithm. The Random Walk-based SFMOs outperform the basic SFMO in all benchmark functions. The results show that predators in SFMO with Random Walk moves in nature-like diversification and intensification that lead to improvement.