Valentin Margraf, M.Sc.
PhD student, Research assistant
Chair of Artificial Intelligence and Machine Learning
Office address:
Akademiestraße 7
Room 122
80799 Munich
Research Focus
My research centers on Bayesian optimization, with a particular focus on making it reliable when its standard assumptions do not hold — such as model misspecification, unknown hyperparameters, or distribution shift. Beyond this, I am also interested in and have worked on active learning, semi-supervised learning and algorithm configuration.
Selected Publications
- Valentin Margraf, Anna Lappe, Marcel Wever, Carolin Benjamins, Eyke Hüllermeier, Marius Lindauer (2025)
SynthACticBench: A Capability-Based Synthetic Benchmark for Algorithm Configuration(bib)(pdf)
Genetic and Evolutionary Computation Conference (GECCO '25), 14. - 18. July, 2025, Malaga, Spain - V Margraf, J Hanselle, J Rodemann, M Wever, S Vollmer, E Hüllermeier (2005)
Imprecise Acquisitions in Bayesian Optimization
Epistemic Intelligence in Machine Learning Workshop at EurIPS 2025