In multi-objective optimization the objective is to reach a point which is Pareto ecient. However we usually encounter many such points and choosing a point amongst them possesses another problem. In many applications we are required to choose a point having a good spread over all objective functions which is a direct consequence of the notion of fairness. In this paper we propose two things. The rst one is the concept of “Feasible Fairness” and the second one is a novel, simple to implement and fast algorithm which is guaranteed to converge to a “feasibly fair” point if the objective functions are continuous and strictly quasi-convex on a compact and convex domain.
Citation
August/2012