TY - GEN
T1 - Revisiting Pareto-Optimal Multi- and Many-Objective Reference Fronts for Continuous Optimization
AU - da Silva, Gabriela Cavalcante
AU - Wanner, Elizabeth F.
AU - Bezerra, Leonardo C.T.
AU - Stützle, Thomas
PY - 2021/8/9
Y1 - 2021/8/9
N2 - The performance assessment of multi-objective heuristic algorithms is one of the most significant contributions from the evolutionary optimization algorithms community. By contrast, performance assessment in the context of many-objective optimization is still a challenging, open research field. Recent advances have demonstrated disagreements between Pareto-compliant performance metrics, and indicated that reference fronts produced by benchmark generators of Pareto-optimal fronts could be further improved. In this work, we investigate these reference fronts with the help of multi-dimensional visualization techniques and Pareto-monotonic archivers. Interestingly, reference fronts produced by benchmark generators for DTLZ and WFG continuous optimization problems show significant issues, even when only three objectives are considered. Furthermore, given that input solution sets for five-objective problems are not high-quality, archivers are unable to output reasonable approximation fronts. We conclude that the performance assessment of EMO algorithms needs to urgently address reference front generation.
AB - The performance assessment of multi-objective heuristic algorithms is one of the most significant contributions from the evolutionary optimization algorithms community. By contrast, performance assessment in the context of many-objective optimization is still a challenging, open research field. Recent advances have demonstrated disagreements between Pareto-compliant performance metrics, and indicated that reference fronts produced by benchmark generators of Pareto-optimal fronts could be further improved. In this work, we investigate these reference fronts with the help of multi-dimensional visualization techniques and Pareto-monotonic archivers. Interestingly, reference fronts produced by benchmark generators for DTLZ and WFG continuous optimization problems show significant issues, even when only three objectives are considered. Furthermore, given that input solution sets for five-objective problems are not high-quality, archivers are unable to output reasonable approximation fronts. We conclude that the performance assessment of EMO algorithms needs to urgently address reference front generation.
UR - https://ieeexplore.ieee.org/document/9504952
UR - http://www.scopus.com/inward/record.url?scp=85124598097&partnerID=8YFLogxK
U2 - 10.1109/CEC45853.2021.9504952
DO - 10.1109/CEC45853.2021.9504952
M3 - Conference publication
AN - SCOPUS:85124598097
T3 - 2021 IEEE Congress on Evolutionary Computation, CEC 2021 - Proceedings
SP - 1171
EP - 1178
BT - 2021 IEEE Congress on Evolutionary Computation, CEC 2021 - Proceedings
PB - IEEE
T2 - 2021 IEEE Congress on Evolutionary Computation, CEC 2021
Y2 - 28 June 2021 through 1 July 2021
ER -