Awesome Differential Privacy

Differential privacy is a mathematical framework for sharing statistical insights about a dataset while provably limiting what can be learned about any individual in it.

Contents

• Code and Projects
• Learn
• Read
• Industry Articles
• YouTube

Code and Projects

• OpenDP - Community effort to build trustworthy, open source tools for differential privacy.
• Google's Differential Privacy Library - C++, Go and Java libraries for computing differentially private statistics.
• Opacus - Train PyTorch models with differential privacy.
• TensorFlow Privacy - Train TensorFlow models with differential privacy.
• Diffprivlib - IBM's general-purpose library of differential privacy mechanisms and machine learning models.
• JAX-Privacy - Algorithms for privacy-preserving machine learning in JAX.
• Harvard University Privacy Tools Project - Research project developing tools and policies for sharing sensitive research data.
• Tumult Analytics - Platform for computing differentially private aggregate statistics on large datasets.
• PyDP - Python wrapper for Google's differential privacy library.
• PipelineDP - Framework for applying differential privacy to large datasets, aimed at non-experts.
• Private Multiplicative Weights - Julia implementation of the MWEM algorithm for differentially private synthetic data.
• diffpriv - R package for easy differential privacy.
• ARX - Data anonymization tool supporting differential privacy, k-anonymity and other privacy models.
• DP-XGBoost - Differentially private gradient boosted trees at scale, built on XGBoost.
• Qrlew - SQL-to-SQL rewriting layer that makes queries differentially private.
• SynthLeak - Experiments on breaking privacy in real-world differentially private synthetic data.
• Fast Differential Privacy - Faster differentially private training of PyTorch models.

Learn

• The Algorithmic Foundations of Differential Privacy - Foundational textbook by Cynthia Dwork and Aaron Roth.
• A Course in Differential Privacy - Video lectures by Gautam Kamath.
• BIU Winter School on Differential Privacy - Lectures from the Bar-Ilan University school "Differential Privacy: From Theory to Practice".
• CS 860 - Algorithms for Private Data Analysis - Graduate course by Gautam Kamath, with lecture notes and videos.
• Secure and Private AI - Udacity course on differential privacy, federated learning and encrypted computation in deep learning.
• A Friendly, Non-Technical Introduction to Differential Privacy - Blog post series by Damien Desfontaines.
• Programming Differential Privacy - Online book about differential privacy, for programmers.
• Differential Privacy in Deep Learning - Introduction to applying differential privacy to deep learning.

Read

• Wikipedia - Encyclopedic overview of differential privacy.
• A Primer for a Non-Technical Audience - Introduction to differential privacy for lawyers and policymakers.
• Differential Privacy for Dummies - Informal introduction by Frank McSherry.
• Introducing TensorFlow Privacy - Announcement of the TensorFlow Privacy library.
• Introducing Opacus - Announcement of the Opacus library.
• An Illustrated Primer - Worked examples of differential privacy by Frank McSherry.
• Win-Vector Blog - Mini-series on differential privacy and reusable holdout sets.
• The Promise of Differential Privacy - Tutorial on algorithmic techniques by Cynthia Dwork.
• Visualizing the Accuracy-Privacy Trade-Off - Interactive demo.
• Laplacian Noisy Counting Mechanism Illustration - Interactive illustration of the Laplace mechanism.

Industry Articles

• Apple - Learning with Privacy at Scale - How Apple uses local differential privacy to learn from user data.
• Apple - Differential Privacy Overview - Apple's overview of its use of differential privacy.
• Google - Deep Learning with Differential Privacy - Paper introducing differentially private stochastic gradient descent.
• Microsoft - How Cynthia Dwork Is Making Data Analysis Better - Profile of the co-inventor of differential privacy.
• Microsoft - Collecting Telemetry Data Privately - How Microsoft collects telemetry with local differential privacy.
• How One of Apple's Key Privacy Safeguards Falls Short - Critique of the privacy parameters used by Apple.
• Google's Differential Privacy May Be Better Than Apple's - Comparison of the two companies' approaches.
• Privacy and Machine Learning: Two Unexpected Allies? - How differential privacy and machine learning can benefit each other.

YouTube

• The Definition of Differential Privacy - Talk by Cynthia Dwork at the Institute for Advanced Study.
• Tutorial on Differential Privacy - Tutorial from the Simons Institute for the Theory of Computing.
• A Practical Beginners' Guide to Differential Privacy - CERIAS security seminar at Purdue University.
• Accessing Data While Preserving Privacy - Talk at the Institut Henri Poincaré.
• Protecting Privacy with MATH - Minutephysics video made in collaboration with the US Census Bureau.

Footnotes