Enhancing invasive weed optimization with taboo strategy

Invasive weed optimization (IWO) is a recently developed metaheuristic that imitates the invasive behavior of weeds in nature. However, the reproduction and spatial dispersal operators in original IWO may make most seeds located around the best weed, which will result in premature convergence. To overcome this drawback, we propose an enhanced IWO algorithm (EIWO) by utilizing the core idea of taboo search. When no better solution is found in the neighborhood of a weed within a certain number of iterations, EIWO judges that this weed has been stagnated and taboos it, thus avoiding the repeated search in its neighborhood. In addition, EIWO also defines a self-production operator which generates some new weeds in a random way rather than directly choosing from the current plant population, so that new solution regions can be explored. To verify the efficiency of the proposed algorithm, we compared it with the original IWO, an improved IWO, and a modified particle swarm optimization on a set of 16 benchmark functions. Computational results indicate that EIWO can prevent premature convergence and produce competitive solutions.