Abstract
Overlapping solutions occur when more than one solution in the space of decisions maps to the same solution in the space of objectives. This situation threatens the exploration capacity of Multi-Objective Evolutionary Algorithms (MOEAs), preventing them from having a good diversity in their population. The influence of overlapping solutions is intensified on multi-objective combinatorial problems with a low number of objectives. This paper presents a hybrid MOEA for handling overlapping solutions that combines the classic NSGA-II with a strategy based on Objective Space Division (OSD). Basically, in each generation of the algorithm, the objective space is divided into several regions using the nadir solution calculated from the current generation solutions. Furthermore, the solutions in each region are classified into non-dominated fronts using different optimization strategies in each of them. This significantly enhances the achieved diversity of the approximate front of non-dominated solutions. The proposed algorithm (called NSGA-II/OSD) is tested on a classic Operations Research problem: the Multi-Objective Knapsack Problem (0-1 MOKP) with two objectives. Classic NSGA-II, MOEA/D and Global WASF-GA are used to compare the performance of NSGA-II/OSD. In the case of MOEA/D two different versions are implemented, each of them with a different strategy for specifying the reference point. These MOEA/D reference point strategies are thoroughly studied and new insights are provided. This paper analyses in depth the impact of overlapping solutions on MOEAs, studying the number of overlapping solutions, the number of solution repairs, the hypervolume metric, the attainment surfaces and the approximation to the real Pareto front, for different sizes of 0-1 MOKPs with two objectives. The proposed method offers very good performance when compared to the classic NSGA-II, MOEA/D and Global WASF-GA algorithms, all of them well-known in the literature.
Highlights
A multi-objective optimization problem (MOP) can be defined as a problem corresponding to a certain reality, where a decision-maker wishes to optimize several objectives simultaneously, usually in conflict with each other
This paper presents a hybrid MultiObjective Evolutionary Algorithms (MOEAs) for handling overlapping solutions that combines the classic Non-dominated Sorting Genetic Algorithm-II (NSGA-II) with a strategy based on Objective Space Division (OSD)
The proposed method offers very good performance when compared to the classic NSGA-II, Multi-Objective Evolutionary Algorithm based on Decomposition (MOEA/D) and Global WASF-GA algorithms, all of them well-known in the literature
Summary
A multi-objective optimization problem (MOP) can be defined as a problem corresponding to a certain reality, where a decision-maker wishes to optimize several objectives simultaneously, usually in conflict with each other. The second generation of Multi-Objective Evolutionary Algorithms (MOEAs) has been shown to achieve excellent results solving MOP [1−4]. The Non-dominated Sorting Genetic Algorithm-II (NSGA-II) [5], the Multi-Objective Evolutionary Algorithm based on Decomposition (MOEA/D) [6] and the Global Weighting Achievement Scalarizing Function Genetic Algorithm (Global WASF-GA) [7], among others, are well-known in the literature. Global WASF-GA, to NSGA-II, uses Pareto dominance as a criterion to converge to the POF but based on the Achievement Scalarizing Function (ASF) proposed by Wierzbicki [8]. The diversity is enhanced by the use of a set of wellspread projection directions, and by not allowing repeating solutions in the same front Both MOEA/D and Global WASF-GA use a scalarizing function as fitness function and simultaneously employ several predefined weight vectors for searching new solutions (aggregation methods).
Sign-in/Register to view highlights & summary 
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call