Handbook of AI-based Metaheuristics: Chapter 2 - Fish School Search: Account for the First Decade

Resumo

Fish School Search (FSS) is a swarm-intelligence subfamily of algorithms proposed by Bastos Filho and Lima Neto in 2008 and first published in 2009. In the FSS, the simple reactive agents are called fish, and each fish has a weight that represents the success obtained during the search. The weights' values and variations influence the individual and collective movements. The embedded mechanisms of feeding and coordinated action make the school move toward the positive gradient to gain weight (and find local and global better positions). Heavier fish have more influence in guiding the search. The idea of accumulating success along the examination indicates that a specific simple reactive agent is worth influencing others. FSS was designed for continuous optimization problems in multimodal search spaces. It has also influenced other researchers to propose variations for other issues, such as optimization in binary problems, multi-objective optimization, many-objective optimization, and multimodal optimization. In this chapter, we present a review of the advances considering FSS in the last decade, including some proposals for binary optimization, three approaches for multi- and many-objective optimizations, and two different multimodal optimization proposals. We also show two other methods for parallel processing, which aim to accelerate the processing time. We finalize the chapter giving some examples of applications of those recent approaches in real-world problems.