Abstract:
|
Data privacy has become a hot-button topic in the aftermath of security breaches and data leaks. Government agencies, businesses, research institutions and social media platforms are being asked to release and share more and more of their data to increase transparency and accountability. However, the simple anonymized methods that have long been used do not provide a high level of privacy. Data intruders can combine their knowledge and other public information to link data sets and identify subjects. In this talk, the notion of differential privacy, a conceptual approach to bring strong mathematical guarantee for privacy protection, will be introduced, along with some of the basic differentially private mechanisms used for adding noise to create synthetic data. Additional work will be presented on the recently developed differentially private data synthesis methods, which use differential privacy while aiming to maintain the analytical validity and utility of the released sanitized data.
|