Multi-Objective Optimization Methods as a Decision Making Strategy

One of the most basic concepts in humankind’s life is the search for an optimal state. As long as the life continues, seeking for perfection in decision making of many areas is fundamental. The concept of optimization is considered as a main goal for a decision maker (DM) to achieve the best solution or most favorable set of solutions of one or more given criteria. In fact, most of real-world problems involve many correlated and often conflicted objectives that should be maximized or minimized to have the problem solved; such setting creates a harder situation for a DM to define all those contradicting objectives in terms of a one single objective following single-objective optimization (SOO) approach. To overcome this limitation, multi-objective optimization (MOO) becomes one of the recent optimization approaches to formulate decision making problems in a more realistic manner. As the ultimate goal of solving MOO problems is to find the optimal set of non-dominated solutions, which is called Paretooptimal set of Pareto-optimal solutions, using classical methods of exact, heuristics or metaheuristics methods become more complicated and cannot guarantee that those optimal solutions will be found. Therefore, many methods have been developed to facilitate the process of solving MOO problems with respect to the role of DM, due to his authority to give further preference information concerning the Pareto-optimal solutions. To this end, this study introduces a brief definition of MOO problem formulation, representation and solution; including analytical comparison of most common MOO problem solving methods in the literature. It’s concluded that MOO methods tree could be classified into four main branches based on DM preference articulation: No preference, A-priori, Interactive, and A-posteriori preferences articulation of DM. Keywords—Multi-objective optimization; No preference; Apriori; A-posteriori; Interactive; Pareto-Optimal; Evolutionary