We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs.LG

Change to browse by:

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Machine Learning

Title: Can Stochastic Gradient Langevin Dynamics Provide Differential Privacy for Deep Learning?

Abstract: Bayesian learning via Stochastic Gradient Langevin Dynamics (SGLD) has been suggested for differentially private learning. While previous research provides differential privacy bounds for SGLD at the initial steps of the algorithm or when close to convergence, the question of what differential privacy guarantees can be made in between remains unanswered. This interim region is of great importance, especially for Bayesian neural networks, as it is hard to guarantee convergence to the posterior. This paper shows that using SGLD might result in unbounded privacy loss for this interim region, even when sampling from the posterior is as differentially private as desired.
Subjects: Machine Learning (cs.LG); Cryptography and Security (cs.CR)
Cite as: arXiv:2110.05057 [cs.LG]
  (or arXiv:2110.05057v4 [cs.LG] for this version)

Submission history

From: Guy Heller [view email]
[v1] Mon, 11 Oct 2021 07:53:54 GMT (89kb,D)
[v2] Tue, 12 Oct 2021 05:37:11 GMT (89kb,D)
[v3] Wed, 2 Feb 2022 06:14:55 GMT (472kb,D)
[v4] Tue, 13 Sep 2022 18:21:05 GMT (347kb,D)

Link back to: arXiv, form interface, contact.