How to Use Heuristics for Differential Privacy

Learn about differential privacy heuristics in this 22-minute IEEE conference talk. Explore the theory, results, and properties of differential privacy, as well as its modular design. Discover the analogy between differential privacy and machine learning, including the learning problem and perfect oracle concept. Examine heuristics such as universal identification, RSPM, and the Wiggle Method. Gain insights into privacy analysis and consider open questions in the field. Presented by Seth Neel, Aaron Roth, and Zhiwei Steven Wu, this talk provides a comprehensive overview of using heuristics for differential privacy applications.

Introduction
Theory
Results
Differential Privacy
Differential Privacy Properties
Modular Design
The Good News
Machine Learning Analogy
Learning Problem
Perfect Oracle
Heuristics
Universal Identification
RSPM
Wiggle Method
Comments
Privacy Analysis
Open Questions