Ideals and Realities of Benchmarking in Evolutionary Multiobjective Optimization. - Tea Tušar
Abstract:
Research in Evolutionary Multiobjective Optimization (EMO) is motivated by real-world problems, which often involve conflicting objectives and complex constraints. Yet, benchmarking in EMO frequently neglects the needs of real-world applications, limiting the relevance and transferability of research findings. This talk will contrast the ideals of benchmarking with the realities observed in current practice. Drawing from insights gained through the study of real-world optimization problems, the presentation will highlight key mismatches between benchmark design and application demands, discuss what current platforms (like COCO) offer, and identify critical gaps that still need to be addressed. The goal is to encourage benchmarking practices that are better aligned with the challenges and diversity of real-world EMO problems.
Tea Tušar is a senior research associate at the Department of Intelligent Systems at the Jožef Stefan Institute and an assistant professor at the Jožef Stefan International Postgraduate School. She received her PhD for her work on visualizing solution sets in multi-objective optimization. Following her doctorate, she completed a postdoctoral fellowship at Inria Lille, France, where she contributed to benchmarking multi-objective optimization algorithms. Her work focuses on Evolutionary Computation, with a particular emphasis on visualizing and benchmarking the results of evolutionary algorithms for both single- and multi-objective optimization, with and without constraints, and applying these methods to solve real-world optimization problems. Tea has been involved in several collaborative projects that apply optimization techniques to practical challenges. These include optimizing electric motor designs, improving energy scheduling systems, and optimizing tunnel alignment. She has contributed to the development of the COCO platform (https://coco-platform.org/) for comparing optimization algorithms, expanding its capabilities to handle multi-objective and mixed-integer problems. Her work continues to focus on advancing optimization tools for both academic research and industrial applications. She has held organizational and editorial roles at PPSN and GECCO and serves as an associate editor for Evolutionary Computation and ACM Transactions on Evolutionary Learning and Optimization. Since 2023, she has also been serving as the Vice President of ACM Slovenia.