Today, Sara Gjorgjieva represented the AutoLearn-SI team at the IEEE Congress on Evolutionary Computation (CEC 2026) in Maastricht, the Netherlands, where she presented our paper "On the Structural (Dis)Agreement of Landscape Representations in Black-Box Optimization." The paper was also nominated for the CEC 2026 Best Student Paper Award.

The research provides one of the first systematic comparisons of modern landscape representations—including ELA, DeepELA, TransOptAS, and DoE2Vec—to understand how they characterize black-box optimization problems. Through extensive clustering, stability, and cross-representation analyses on the MA-BBOB benchmark suite, the study demonstrates that different representations capture fundamentally different structures of the same optimization landscape, highlighting important implications for automated algorithm selection, meta-learning, and trustworthy benchmarking.

The work is a collaboration between Sara Gjorgjieva, Eva Tuba, Barbara Koroušić Seljak, Carola Doerr, and Tome Eftimov.

We congratulate Sara on an excellent presentation and wish her the best of luck in the Best Student Paper Award competition!