Sample-and-Threshold Differential Privacy: Histograms and Applications
Abstract
Federated analytics aims to compute accurate statistics from distributed datasets. A "Differential Privacy" (DP) guarantee is usually desired by the users of the devices storing the data. In this work, we prove a strong $(\epsilon, \delta)$-DP guarantee for a highly practical sampling-based procedure to derive histograms. We also provide accuracy guarantees and show how to apply the procedure to estimate quantiles and modes.
Cite
Text
Bharadwaj and Cormode. "Sample-and-Threshold Differential Privacy: Histograms and Applications." NeurIPS 2021 Workshops: PRIML, 2021.Markdown
[Bharadwaj and Cormode. "Sample-and-Threshold Differential Privacy: Histograms and Applications." NeurIPS 2021 Workshops: PRIML, 2021.](https://mlanthology.org/neuripsw/2021/bharadwaj2021neuripsw-sampleandthreshold/)BibTeX
@inproceedings{bharadwaj2021neuripsw-sampleandthreshold,
title = {{Sample-and-Threshold Differential Privacy: Histograms and Applications}},
author = {Bharadwaj, Akash and Cormode, Graham},
booktitle = {NeurIPS 2021 Workshops: PRIML},
year = {2021},
url = {https://mlanthology.org/neuripsw/2021/bharadwaj2021neuripsw-sampleandthreshold/}
}