Multi-objective Optimization and its Pareto Extension

Abstract

In real life optimization problems, we often seek solutions representing the best trade-offs between conflicting objectives. Existing methods dealing with multi-objective optimization usually output a solution representing a single pre-defined trade-off. In order to produce additional meaningful trade-offs, we present Diversity Maximization Algorithm (DMA) for Multi-objective optimization. This algorithm outputs a set of diverse optimal solutions that lie on Pareto Frontier, thus allowing the user to efficiently explore the optimal solutions space.