Optimizing multi-plateau functions with FSS-SAR (Stagnation Avoidance Routine)
Joao Batista Monteiro Filho, Isabela Albuquerque, Fernando Buarque de Lima Neto, Filipe Vieira Silva Ferreira · 2016
In this article we investigate the effectiveness of a stagnation avoidance routine devised for Fish School Search algorithm, which is a novel nature inspired population based search procedure effective for continuous optimization problems. Here we used and modified either Vanilla and a niching version of the algorithm in order to improve their exploratory ability with the introduction of a stochastic worsening allowance behavior within the local search operator, originating the stagnation avoidance routine. Two sets of multi-plateau objective functions were defined in order to evaluate the performance improvement in multi-plateau search spaces, a common and challenging threat for Combinatorial Optimization algorithms. The main goal was to improve the convergence capability of the algorithm when solving very smooth or plateau containing search spaces. Instances of the multi-plateau functions as well as a set of benchmark test problems were solved. Results show that the proposed modification for Fish School Search is quite effective in abbreviating as well as improving the convergence rate of the algorithm.