A Comparative Study Use of OTL for Many-objective Optimization
Jinhua Zheng, Hui Bai, Ruimin Shen, Miqing Li · 2015
This study exhaustively compares the abilities to solve many-objective problems of eight representative algorithms from four different classes (i.e., Pareto-, aggregation-, indicator-, and diversity-based EMO algorithms). The eight compared algorithms are tested on four types of well-defined continuous, discontinuous and combinatorial problems, through three performance metrics as well as a visual observation in the decision space. We can conclude from the experimental results that the performance of the eight algorithms differ not only on the dimensionality of the problems, but also on the shape and features of the Pareto front. From this it suggests an appropriate choice for researchers and practitioners when solving many-objective problems.