A New Model for International, Privacy-Preserving Data Science

Explore a groundbreaking approach to international, privacy-preserving data science in this 11-minute conference talk from PEPR '24. Delve into Curtis Mitchell's presentation from xD and the US Census Bureau, which introduces a novel model for conducting data analysis between National Statistical Organizations (NSOs). Learn how this innovative method utilizes remote, privacy-preserving processes to overcome traditional restrictions and complexities in data sharing. Discover the collaboration between multiple NSOs and the United Nations Privacy-Enhancing Technologies Lab (UN PET Lab), and understand the proof-of-concept implementation using the open-source PySyft platform. Gain insights into the cloud infrastructure setup, including nodes hosted by the US Census Bureau and other NSOs, facilitated by a UN PET Lab network gateway. Explore how this architecture enables private joins on synthetic data representing UN Comtrade trade information without direct data access between NSOs. Examine the potential impact of this project on future privacy policy and governance in international data science collaborations.