User-Level Differential Privacy in Medical Machine Learning
MCML Authors
Daniel Rückert
Prof. Dr.
Director
Abstract
Daniel Rückert
Prof. Dr.
Director
Abstract
We address the challenge of ensuring user-level DP when individuals contribute varying numbers of data records to a dataset. While group privacy can be used to aggregate record-level budgets, it can be overly pessimistic and lacks flexibility when users contribute varying numbers of data points. We propose a method for accounting for arbitrary numbers of records per user while maintaining a fixed per-user privacy guarantee by leveraging individual privacy assignment. Experimentally, our method yields excellent utility comparable to record-level DP while providing a more meaningful/interpretable protection.
inproceedings KER+25
TPDP 2025
Workshop on Theory and Practice of Differential Privacy. Google, Mountain View, CA, USA, Jun 02-03, 2025.Authors
J. Kaiser • J. Eigenmann • D. Rückert • G. KaissisLinks
PDFResearch Area
BibTeXKey: KER+25