Algorithms for Multi-objective Optimization

Many practical applications are better modeled as optimization problem, characterized by the existence of multiple conflicting objectives. A classical and usual example is the compromise between maximizing consumer satisfaction and minimizing service cost. Indeed, dealing with conflicting objectives is omnipresent in our lives, and a significant portion of these multi-objective problems admits a proper mathematical formulation, so that we may resort to computational resources to obtain Pareto-optimal solutions, also called non-inferior solutions.

Without any assumption from the decision maker, all non-inferior solutions are equally relevant, so what we want to research in this project are algorithms capable of find a good representative set of non inferior solutions - those methods are called a posteriori multiobjective methods. This optimization framework can be applied to logistic problems, economics and machine learning.