COCO (Comparing Continuous Optimizers) is an open-source platform for benchmarking single-objective and multi-objective optimizers on a black-box setting. It facilitates benchmarking of optimization algorithms by automatizing this procedure and providing data of previously run algorithms for comparison.
F.12 Improvements to an existing service
COBISS.SI-ID: 33003303COCO (Comparing Continuous Optimizers) is an open-source platform for benchmarking single-objective and multi-objective optimizers on a black-box setting. It facilitates benchmarking of optimization algorithms by automatizing this procedure and providing data of previously run algorithms for comparison.
F.12 Improvements to an existing service
COBISS.SI-ID: 31245095This lecture gives an overview of the status of benchmarking multiobjective optimizers and proposes ways to improve it, specifically by using real-world problems in addition to of artificial ones when benchmarking algorithms. After presenting the drawbacks of the current benchmarking methodology, we look at the anytime performance assessment approach supported by the COCO platform. We then list existing real-world problems that could be used for testing optimization algorithms and discuss how they could be integrated into COCO.
B.04 Guest lecture
COBISS.SI-ID: 31453479This lecture introduces students to multiobjective optimization with evolutionary algorithms. It presents the basic concepts of multiobjective optimization, the challenges of comparing non-dominated solution sets and three different approaches to solving multiobjective problems. It uses examples to show the characteristics of real-world optimization problems and present the mechanisms of evolutionary algorithms, which make it possible to solve such problems efficiently. The lecture concludes with the presentation of additional challenges of multiobjective optimization, such as many objectives, problems with constraints and expensive solution evaluations.
B.04 Guest lecture
COBISS.SI-ID: 30943015An interview for the Gea magazine, where we discuss the use of evolutionary algorithms for solving multiobjective optimization as well as women in computer science.
F.35 Other
COBISS.SI-ID: 31051047