A new genetic operator to improve the diversity of the Multiobjective Evolutionary Algorithms

The optimization of many objectives requires a set of optimal solutions known as Pareto solutions. Similarly to the optimization of single objective in Evolutionary Algorithms (EAs), the Multiobjective Evolutionary Algorithms (MOEAs) also suffer from loss of genetic diversity, allowing the appearance of sparse regions along the Pareto frontier. A mechanism to maintain the population diversity along generations is needed. It is expected that, if diversity is controlled effectively, at the end of the evolutionary process, the Pareto Front optimum will be as uniformly distributed as possible. This paper proposes a new diversity operator that generates artificial solutions to fill sparse regions of the non-dominated set of solutions found by the MOEA. It uses artificial neural networks (ANN) to perform a reverse mapping from the phenotype to the corresponding genotype of an inserted artificial solution. This mechanism was tested with NSGA-II and SPEA2 algorithms. The addition of the diversity operator reached significant improvements in the hypervolume and the spread metrics of the obtained set of solutions non-dominated.