A tree-of-heaven inspired approach for solving many-objective optimization problems

Evolutionary algorithms face some challenges dealing with problems with more than three objectives. In particular, the crossover operator loses its efficiency due to the long distance between solutions. In addition, the selection pressure of domination-based approaches decreases because almost all of the population becomes non-dominated. Moreover, the diversity evaluations become too costly. In this paper, we mimic the behavior of Ailanthus altissima, commonly known as tree-of-heaven, in maximum use of natural resources to establish a balance between convergence and diversity of the population. This invasive tree spreads rapidly by adapting to changes in light. More specifically, on the one hand, in order to maintain territory it directs its branches towards the sun (the ideal point) in the more crowded regions; on the other hand, it first tries to expand territory (diversity) in less crowded regions and grow taller afterwards. Moreover, in addition to seeding strategy it uses root sprouts to dominate colonized sites. The proposed algorithm, called THIA, is compared with six of the state-of-the-art algorithms on four unconstrained many-objective test problems with up to 15 objectives. The results show that THIA works well on almost all of the selected many-objective problems.