Hindrances for robust multi-objective test problems
Affiliation:
1. School of Information and Communication Technology, Griffith University, Nathan, Brisbane, QLD 4111, Australia;2. Queensland Institute of Business and Technology, Mt Gravatt, Brisbane QLD 4122, Australia;1. Department of Mechanical Engineering, Universiti Teknologi Petronas, Malaysia;2. Department Civil Engineering, Universiti Teknologi Petronas, Malaysia;1. Department of Information Systems Management and GREC Group, Esade – Universitat Ramon Llull, Barcelona, Spain;2. Department of Neonatology, Máxima Medical Center, The Netherlands;3. Department of Applied Mathematics 2 and GREC Group, Technical University of Catalonia, UPC-BarcelonaTech, Barcelona, Spain;1. Departamento de Lenguajes y Sistemas Informáticos, Universidad de Sevilla, Avd. Reina Mercedes s/n, 41012 Seville, Spain;2. Department of Computer Science, School of Engineering, Pablo de Olavide University, Ctra. Utrera km. 1, 41013 Seville, Spain;1. Information Systems Department, ESADE Business School, Ramon Llull University, Av. Pedralbes, 60-62, Barcelona, Spain;2. Data Science and Operations Department, Marshall School of Business, University of Southern California, Los Angeles, USA;1. School of Electrical & Automatic Engineering, Changshu Institute of Technology, 215500 Changshu, China;2. School of Automation, Nanjing University of Science & Technology, 210094 Nanjing, China
Abstract:
Despite the significant number of benchmark problems for evolutionary multi-objective optimisation algorithms, there are few in the field of robust multi-objective optimisation. This paper investigates the characteristics of the existing robust multi-objective test problems and identifies the current gaps in the literature. It is observed that the majority of the current test problems suffer from simplicity, so five hindrances are introduced to resolve this issue: bias towards non-robust regions, deceptive global non-robust fronts, multiple non-robust fronts (multi-modal search space), non-improving (flat) search spaces, and different shapes for both robust and non-robust Pareto optimal fronts. A set of 12 test functions are proposed by the combination of hindrances as challenging test beds for robust multi-objective algorithms. The paper also considers the comparison of five robust multi-objective algorithms on the proposed test problems. The results show that the proposed test functions are able to provide very challenging test beds for effectively comparing robust multi-objective optimisation algorithms. Note that the source codes of the proposed test functions are publicly available at www.alimirjalili.com/RO.html.
