HPO, Interim Professor, Postdoctoral Research Fellow
Steven Adriaensen
Postal address
Institut für InformatikAlbert-Ludwigs-Universität Freiburg
Sekretariat Hutter/Maschinelles Lernen
Georges-Köhler-Allee 074
79110 Freiburg, Germany
Office
Building 074, Room 00-015
About
Since June 2024, I serve as Interim Professor for Machine Learning at the University of Freiburg, where I lead the AutoML group during Frank Hutter’s leave. Before that, I was a postdoctoral researcher in the same group and obtained my PhD in Computer Science at the Vrije Universiteit Brussel, Belgium. My work bridges algorithmics and machine learning, with a focus on automating the design, selection, and configuration of algorithms.
Research Interests
My research focuses on improving algorithmic decision-making, which in practice still often depends on ad-hoc intuition and trial-and-error. I aim to develop methods that both support human experts in designing better algorithms and automate the selection and configuration process.
Most recently, my work has concentrated on Prior-data Fitted Networks (PFNs), models pretrained for in-context learning. This paradigm is emerging as a highly promising direction for algorithm performance prediction. I investigate how PFNs can serve as surrogate models for grey-box hyperparameter optimization (HPO), improving sample efficiency and enabling scaling to modern deep learning.
More broadly, my research spans:
-
In-Context Learning and Prior-data Fitted Networks (PFNs)
-
Grey-box Hyperparameter Optimization
-
Automated Machine Learning (AutoML)
-
Dynamic Algorithm Configuration (DAC)
-
Learning to Learn / Meta-Learning
-
Deep Learning and Optimization
Publications
2025 |
Gompertz Linear Units: Leveraging Asymmetry for Enhanced Learning Dynamics Inproceedings Forthcoming In: 39th Conference on Neural Information Processing Systems (NeurIPS), Forthcoming. |
Bayesian Neural Scaling Laws Extrapolation with Prior-Fitted Networks Inproceedings In: Proceedings of the 42nd International Conference on Machine Learning (ICML), 2025. |
$alpha$-PFN: In-Context Learning Entropy Search Inproceedings In: The Frontiers in Probabilistic Inference: Sampling meets Learning (FPI) at ICLR, 2025. |
2024 |
NOSBench-101: Towards Reproducible Neural Optimizer Search Inproceedings In: Proceedings of the Third International Conference on Automated Machine Learning (AutoML 2024), Workshop Track, 2024. |
From Epoch to Sample Size: Developing New Data-driven Priors for Learning Curve Prior-Fitted Networks Inproceedings In: Proceedings of the Third International Conference on Automated Machine Learning (AutoML 2024), Workshop Track, 2024. |
In-Context Freeze-Thaw Bayesian Optimization for Hyperparameter Optimization Inproceedings In: Proceedings of the Third International Conference on Automated Machine Learning (AutoML 2024), Workshop Track, 2024. |
In-Context Freeze-Thaw Bayesian Optimization for Hyperparameter Optimization Inproceedings In: Proceedings of the 41st International Conference on Machine Learning (ICML), 2024. |
2023 |
Efficient Bayesian Learning Curve Extrapolation using Prior-Data Fitted Networks Inproceedings In: Thirty-seventh Conference on Neural Information Processing Systems (NeurIPS), 2023. |
2022 |
Automated Dynamic Algorithm Configuration Journal Article In: Journal of Artificial Intelligence Research (JAIR), vol. 75, pp. 1633-1699, 2022. |
Efficient Bayesian Learning Curve Extrapolation using Prior-Data Fitted Networks Inproceedings In: Sixth Workshop on Meta-Learning at the Conference on Neural Information Processing Systems, 2022. |
2021 |
DACBench: A Benchmark Library for Dynamic Algorithm Configuration Inproceedings In: Proceedings of the Thirtieth International Joint Conference on Artificial Intelligence (IJCAI'21), ijcai.org, 2021. |
2020 |
Learning Step-Size Adaptation in CMA-ES Inproceedings In: Proceedings of the Sixteenth International Conference on Parallel Problem Solving from Nature (PPSN'20), 2020. |