Factorization by extremal privacy mechanisms: new insights into efficiency
Chiara Amorino, Arnaud Gloter · 2025
Read Factorization by extremal privacy mechanisms: new insights into efficiency by Chiara Amorino, Arnaud Gloter (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2507-21769.txt.
Description
Chiara Amorino and Arnaud Gloter study how well private data can still inform estimates in a 2025 math paper. The setting is local differential privacy, where each person's data is scrambled before it leaves their hands. Stronger scrambling protects people but drains the signal, and the paper asks how much signal can survive.
A splitting lemma does the heavy lifting: any scrambling scheme breaks into an extreme scheme plus extra noise. That fact shrinks the search for the best scheme to extreme ones alone. The authors prove matching upper and lower bounds on the best achievable information in the strong-privacy limit. They also show a best scheme always exists. A worked case, estimating the range of a uniform spread, yields an efficient estimator.
Catalog record
| CREATOR | Chiara Amorino, Arnaud Gloter |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 160 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2507-21769