Topics in Data Privacy

Catalog description

Examines modern threats to data privacy and state-of-the-art approaches for protecting sensitive information. The first half highlights the limitations of traditional anonymization techniques and surveys contemporary privacy-preserving methods, including cryptographic-based approaches such as secure multi-party computation, zero-knowledge proofs, and homomorphic encryption. The second half focuses on differential privacy, the current standard for rigorous privacy guarantees. Topics include core definitions, composition, and fundamental mechanisms, as well as the local model used in real-world data collection. Time permitting, advanced applications such as synthetic data generation and privacy-preserving machine learning will be covered. Coursework includes implementing privacy algorithms and conducting experiments using popular languages for data science such as Python and Julia. No prior background is required. Prerequisites: Familiarity with Python programming and CS 375 and either MATH 327 or MATH 448 or equivalent. Expected to be offered once every academic year.