PLD accountant
Project description
PLD-Accountant
Python code for computing exact DP-guarantees for the subsampled Gaussian mechanism.
The method is described in:
Antti Koskela, Joonas Jälkö, Antti Honkela:
Computing Exact Guarantees for Differential Privacy
https://arxiv.org/abs/1906.03049
Usage
You can download a PyPI package
pip3 install pld-accountant
and run, for example,
from pld_accountant import compute_eps,compute_delta
q=0.01
sigma=1.2
nc=1000 #number of compositions
delta=1e-5
a = compute_eps.get_epsilon_bounded(q=q,sigma=sigma,target_delta=delta,ncomp=nc)
eps=2.0
d = compute_delta.get_delta_bounded(q=q,sigma=sigma,target_eps=eps,ncomp=nc)
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
pld_accountant-0.12.0.tar.gz
(8.3 kB
view hashes)
Built Distribution
Close
Hashes for pld_accountant-0.12.0-py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | 5edfa8136240ffb12ccab7071eb74fb603a4f57dc0a89a77baa39ceb4de8ed4e |
|
MD5 | 5b94935d1ebc616d3118a8e8f51398db |
|
BLAKE2b-256 | 19f82c8b36baed05ba9de268e57446bdd80d4d88fe8285eb1316760662692b05 |