Abstract

Privacy Enhancing Technologies (PETs) are an emerging class of technologies with a promise to protect the privacy and confidentiality of data throughout its life cycle, while maintaining its utility. PETs provide Statistical Offices opportunities to facilitate collaborative analytic on less-accessible data to derive valuable insights. Statistics Canada has started experimenting with PETs a few years ago. To this end, multiple research projects have successfully been completed, such as the application of homomorphic encryption on training a machine learning (ML) classifier, privacy preserving record linkage with secure Multi-Party Computation and applying Federated Learning in the context of privacy preserving crowdsourcing. In this contribution, we will discuss some of these activities and share insights on potential opportunities and challenges of adopting PETs in the Official Statistics.