Multi-objective Optimization Through Coevolution and Outranking Methods with Uncertainty Management

Multi-objective optimization has evolved significantly to this day, but there are still challenges that remain open problems such as algorithmic design, scalability and handling of costly objective functions. To solve some of the aforementioned challenges, in the state of the art it is being proposed to apply coevolutionary algorithms and thus solve large-scale problems. Other authors propose applying outranking methods and uncertainty, which help when deciding to find solutions according to preferences, to direct the search towards the region of interest. To our knowledge, it was found that we have worked with multi-objective optimization with uncertainty or with coevolution, but we did not find works where they are present in a combined way. Therefore, in this work, it is proposed to develop a coevolutionary algorithm that uses outranking relationships to solve multi-objective optimization problems with uncertainty and that, in addition, attends to the scalability in the decision variables and in the number of objective functions.