PET Talk 9: Deployable Differential Privacy

Beschreibung

Differential privacy (DP) is a mathematical framework for protecting individual privacy in computations on sensitive data. Despite robust privacy guarantees and growing institutional support, DP has seen limited adoption. Research has identified multiple barriers, such as privacy risk assessment challenges, utility concerns, and communication difficulties across diverse stakeholders. Overall, fewer than a dozen organizations have publicly deployed DP, predominantly US government agencies and big tech companies.

In his talk, Shlomi Hod will propose a vision for deployable differential privacy, grounded in his own research and that of others. As the demand for "microdata format" releases increases, the talk will focus on differentially private synthetic data. This vision will be demonstrated through a national birth registry release Shlomi leads, which represents the first differentially private release of medical data, developed through a co-design process with diverse stakeholders and which tackles these barriers. He'll conclude the talk by presenting a future research agenda to advance the deployability of differential privacy.