Sequential Auditing for f-Differential Privacy
Tim Kutta, Martin Dunsche, Yu Wei, Vassilis Zikas · 2026
Read Sequential Auditing for f-Differential Privacy by Tim Kutta, Martin Dunsche, Yu Wei, Vassilis Zikas (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-06518.txt.
Description
Checking that an algorithm truly protects privacy, rather than just claiming to, is the job of empirical privacy auditors. Tim Kutta and 3 co-authors build new auditors for f-differential privacy, a richer notion that captures the whole privacy trade-off curve instead of a single pair of numbers. Most existing auditors work in batches and target the older, simpler notion.
Their auditors spot violations across the whole privacy range with statistical guarantees, backed by theory and simulations. Unlike prior tools, they need no user-supplied sample size: they adaptively find a near-optimal number of samples to reach a decision. That thrift matters most for expensive training runs like differentially private gradient descent. Both white-box and black-box settings are supported.
Catalog record
| CREATOR | Tim Kutta, Martin Dunsche, Yu Wei, Vassilis Zikas |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 73 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602-06518