DeepAI AI Chat
Log In Sign Up

FriendlyCore: Practical Differentially Private Aggregation

10/19/2021
by   Eliad Tsfadia, et al.
Google
cohenwang.com
Tel Aviv University
SEO-URI
10

Differentially private algorithms for common metric aggregation tasks, such as clustering or averaging, often have limited practicality due to their complexity or a large number of data points that is required for accurate results. We propose a simple and practical tool 𝖥𝗋𝗂𝖾𝗇𝖽𝗅𝗒𝖢𝗈𝗋𝖾 that takes a set of points D from an unrestricted (pseudo) metric space as input. When D has effective diameter r, 𝖥𝗋𝗂𝖾𝗇𝖽𝗅𝗒𝖢𝗈𝗋𝖾 returns a "stable" subset D_G⊆ D that includes all points, except possibly few outliers, and is certified to have diameter r. 𝖥𝗋𝗂𝖾𝗇𝖽𝗅𝗒𝖢𝗈𝗋𝖾 can be used to preprocess the input before privately aggregating it, potentially simplifying the aggregation or boosting its accuracy. Surprisingly, 𝖥𝗋𝗂𝖾𝗇𝖽𝗅𝗒𝖢𝗈𝗋𝖾 is light-weight with no dependence on the dimension. We empirically demonstrate its advantages in boosting the accuracy of mean estimation, outperforming tailored methods.

READ FULL TEXT

page 1

page 2

page 3

page 4

06/05/2020

Differentially private partition selection

Many data analysis operations can be expressed as a GROUP BY query on an...
02/16/2021

Differentially Private Quantiles

Quantiles are often used for summarizing and understanding data. If that...
12/29/2021

Differentially-Private Clustering of Easy Instances

Clustering is a fundamental problem in data analysis. In differentially ...
10/30/2018

Private Algorithms Can Always Be Extended

We consider the following fundamental question on ϵ-differential privacy...
10/30/2018

Private Algorithms Can be Always Extended

We consider the following fundamental question on ϵ-differential privacy...
11/12/2018

Boosting Model Performance through Differentially Private Model Aggregation

A key factor in developing high performing machine learning models is th...
07/19/2016

Nested Kriging predictions for datasets with large number of observations

This work falls within the context of predicting the value of a real fun...