Metraipolis › archive › dc_511ce7e55523

Revisiting the Provable-Auditable Privacy Gap of DP-SGD — arXiv:2608.28934

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.28934
License (as archived)
CC0-1.0
Archived
2026-10-17T08:32:31.406Z
SHA-256 of the archived text
3215fa8e32ef6b74e2b7004d985bab73b6b0886e789f0912509ad9d435b591fc
Stable URL of this page
https://metraipolis.com/documento/dc_511ce7e55523

How to cite

Revisiting the Provable-Auditable Privacy Gap of DP-SGD — arXiv:2608.28934 — arXiv (en), https://arxiv.org/abs/2608.28934 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_511ce7e55523 (Metraipolis archive dc_511ce7e55523, sha256 3215fa8e32ef6b74e2b7004d985bab73b6b0886e789f0912509ad9d435b591fc)
BibTeX
@misc{dc511ce7e55523,
  title = {Revisiting the Provable-Auditable Privacy Gap of DP-SGD — arXiv:2608.28934},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.28934},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_511ce7e55523; sha256 3215fa8e32ef6b74e2b7004d985bab73b6b0886e789f0912509ad9d435b591fc},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.28934. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Differential privacy (DP) has traditionally been used to provide theoretical upper bounds on an algorithm's stability to changing its training data. In modern private machine learning applications, achieving strong tradeoffs between utility and theoretical privacy is challenging, and thus one may optimistically hope that existing theoretical privacy analyses are loose. Recent work on privacy auditing has adopted a dual viewpoint, instead lower bounding the true privacy of an algorithm by constructing empirical distinguishing events. The auditing literature has thus far yielded a pessimistic outlook on the looseness of theoretical privacy bounds for DP-SGD, the de facto private training method in modern ML, as nearly-matching empirical lower bounds have been achieved under various threat models [NHSBTJCT23, AC24, CBP25]. In this work, we propose the empirical privacy lower bound of an algorithm as a concrete metric to optimize for, complementary to the theoretical upper bound. We give a lightweight defense framework that generically augments optimization methods in the ML pipeline to have significantly-improved empirical privacy on standard benchmarks. Moreover, we show that our framework comes at no theoretical privacy cost when augmenting DP-SGD, unlike previously-proposed defenses against membership inference attacks. We evaluate our defense against a broad range of audit constructions, models, and datasets to demonstrate its flexibility.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-01 in cs.LG).
Categorie arXiv: Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2608.28934 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-17T08:13:26.715Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-09-01?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).