From Scores to Decisions
A short note on turning anomaly scores into calibrated evidence and keeping the decision layer inspectable.
- conformal inference
- anomaly detection
- online FDR
Research staff / Machine learning / Uncertainty
Oliver Hennhöfer
I focus on ML systems that make uncertainty explicit, especially through conformal inference.
Research Focus
My work sits at the intersection of inferential statistics and applied machine learning.
- Uncertainty Quantification. Conformal inference methods for trustworthy decision-making.
- Anomaly Detection. Unsupervised methods for identifying rare patterns in ill-posed settings.
Projects
nonconform
Python package for calibrated anomaly detection and decision-aware conformal inference methods.
online-fdr
Online false discovery rate and sequential decision-control methods for hypothesis testing streams.
aberrant
Anomaly detection experiments and streaming workflows for reproducible baseline comparisons.
Publications
2026