Sequentially Auditing Differential Privacy
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866914158182662144 |
|---|---|
| author | González, Tomás Dulce-Rubio, Mateo Ramdas, Aaditya Ribero, Mónica |
| author_facet | González, Tomás Dulce-Rubio, Mateo Ramdas, Aaditya Ribero, Mónica |
| contents | We propose a practical sequential test for auditing differential privacy guarantees of black-box mechanisms. The test processes streams of mechanisms' outputs providing anytime-valid inference while controlling Type I error, overcoming the fixed sample size limitation of previous batch auditing methods. Experiments show this test detects violations with sample sizes that are orders of magnitude smaller than existing methods, reducing this number from 50K to a few hundred examples, across diverse realistic mechanisms. Notably, it identifies DP-SGD privacy violations in \textit{under} one training run, unlike prior methods needing full model training. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2509_07055 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Sequentially Auditing Differential Privacy González, Tomás Dulce-Rubio, Mateo Ramdas, Aaditya Ribero, Mónica Cryptography and Security Machine Learning Methodology We propose a practical sequential test for auditing differential privacy guarantees of black-box mechanisms. The test processes streams of mechanisms' outputs providing anytime-valid inference while controlling Type I error, overcoming the fixed sample size limitation of previous batch auditing methods. Experiments show this test detects violations with sample sizes that are orders of magnitude smaller than existing methods, reducing this number from 50K to a few hundred examples, across diverse realistic mechanisms. Notably, it identifies DP-SGD privacy violations in \textit{under} one training run, unlike prior methods needing full model training. |
| title | Sequentially Auditing Differential Privacy |
| topic | Cryptography and Security Machine Learning Methodology |
| url | https://arxiv.org/abs/2509.07055 |