Deep Statistical Comparison in Multi-Objective Optimization

AbstractThis chapter presents an application of the Deep Statistical Comparison approach in multi-objective optimization. It provides examples of how the Deep Statistical Comparison ranking scheme can be used for performance assessment of multi-objective stochastic optimization algorithms using a single-quality-indicator data. Next, different ensembles of quality indicators based on the Deep Statistical Comparison ranking scheme are introduced to reduce the user preference bias in the selection of the quality indicators that are involved in the performance assessment. Finally, a version of the Deep Statistical Comparison ranking scheme for handling high-dimensional data as Pareto fronts is introduced together with its application in benchmarking studies.