Diversity Assessment of Multi-Objective Evolutionary Algorithms: Performance Metric and Benchmark Problems [Research Frontier]

Abstract

Diversity preservation plays an important role in the design of multi-objective evolutionary algorithms, but the diversity performance assessment of these algorithms remains challenging. To address this issue, this paper proposes a performance metric and a multi-objective test suite for the diversity assessment of multiobjective evolutionary algorithms. The proposed metric assesses both the evenness and spread of a solution set by projecting it to a lower-dimensional hypercube and calculating the "volume" of the projected solution set. The proposed test suite contains eight benchmark problems, which pose stiff challenges for existing algorithms to obtain a diverse solution set. Experimental studies demonstrate that the proposed metric can assess the diversity of a solution set more precisely than existing ones, and the proposed test suite can be used to effectively distinguish between algorithms with respect to their diversity performance.

Publication
IEEE Computational Intelligence Magazine

Related