An Evolutionary Algorithm based on an outranking relation for sorting problems

A new approach for using the preferences elicited from a Decision Maker (DM) into the operational framework of an Evolutionary Algorithm (EA) is presented. The preference representation is achieved using the parameters and principles of the ELECTRE TRI method devoted to the sorting problem. The outranking relation is used to replace the non-dominance relation in the usual operators in the EA (crossover, mutation and selection operator). The aim of this approach is to focus the search on the region of interest defined by the DM´s preferences and consequently restrict the number of solutions in the Pareto-optimal front to be subject to further screening. This aspect is particularly important when dealing with problems that lead to a large number of non-dominated solutions.