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

Computer Science > Cryptography and Security

Title: Optimal Compression of Locally Differentially Private Mechanisms

Abstract: Compressing the output of \epsilon-locally differentially private (LDP) randomizers naively leads to suboptimal utility. In this work, we demonstrate the benefits of using schemes that jointly compress and privatize the data using shared randomness. In particular, we investigate a family of schemes based on Minimal Random Coding (Havasi et al., 2019) and prove that they offer optimal privacy-accuracy-communication tradeoffs. Our theoretical and empirical findings show that our approach can compress PrivUnit (Bhowmick et al., 2018) and Subset Selection (Ye et al., 2018), the best known LDP algorithms for mean and frequency estimation, to to the order of \epsilon-bits of communication while preserving their privacy and accuracy guarantees.
Subjects: Cryptography and Security (cs.CR); Machine Learning (cs.LG)
Cite as: arXiv:2111.00092 [cs.CR]
  (or arXiv:2111.00092v1 [cs.CR] for this version)

Submission history

From: Abhin Shah [view email]
[v1] Fri, 29 Oct 2021 21:36:34 GMT (1412kb,D)
[v2] Sat, 26 Feb 2022 17:56:55 GMT (1416kb,D)

Link back to: arXiv, form interface, contact.