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

Download:

Current browse context:

cs.CR

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 > Cryptography and Security

Title: Kantorovich Mechanism for Pufferfish Privacy

Authors: Ni Ding
Abstract: Pufferfish privacy achieves $\epsilon$-indistinguishability over a set of secret pairs in the disclosed data. This paper studies how to attain $\epsilon$-pufferfish privacy by exponential mechanism, an additive noise scheme that generalizes the Laplace noise. It is shown that the disclosed data is $\epsilon$-pufferfish private if the noise is calibrated to the sensitivity of the Kantorovich optimal transport plan. Such a plan can be obtained directly from the data statistics conditioned on the secret, the prior knowledge of the system. The sufficient condition is further relaxed to reduce the noise power. It is also proved that the Gaussian mechanism based on the Kantorovich approach attains the $\delta$-approximation of $\epsilon$-pufferfish privacy.
Comments: 20 pages, incl. supplementary materials, 4 figures, to appear in proceeding of AISTATS 2022
Subjects: Cryptography and Security (cs.CR); Applications (stat.AP)
Cite as: arXiv:2201.07388 [cs.CR]
  (or arXiv:2201.07388v2 [cs.CR] for this version)

Submission history

From: Ni Ding Dr [view email]
[v1] Wed, 19 Jan 2022 02:30:41 GMT (1968kb)
[v2] Mon, 21 Feb 2022 05:13:14 GMT (1248kb,D)

Link back to: arXiv, form interface, contact.