Overview
| The state of the art in Privacy-Preserving Machine Learning (PPML) is promising from a technical perspective but under-explored if we view it through the lens of legal instruments such as the EU AI Act and the EU Data Act. In this roundtable, we aim to contribute to clearing the path to alternative solutions for processing (personal) data in Machine Learning applications. In addition, how and why data is used has become more complex and can no longer be discussed in a binary way. Innovation happens in a myriad of ways. Think, e.g., the use of machine learning to develop new products or to innovate responsibly in that sphere. Machine learning can also be used to advance societal goals or to protect individuals’ fundamental rights. It will be key to take such data needs and technological advancements into account in ongoing and future regulatory discussions on, for example, the EU AI Act, the European Health Data Space (EHDS) Regulation, or a future revision of the GDPR. Date: 8 December 2022 Time: from 15:00 PM – 17:00 PM CET Location: Microsoft Center, Rue Montoyer 51, Brussels 1000, Bruxelles Agenda: 15:00-15:10 Introductory comments by Rob van Eijk (Managing Director for Europe, Future of Privacy Forum) and Gianclaudio Malgieri (Co-Director of the Brussels Privacy Hub & Associate Professor of Law at Leiden University) 15:10-15:23 Overview of anonymization, deidentification landscape by Andrea Gadotti (University of Oxford) 15:23-15:33 Differential privacy (in particular DP-SGD) + measurements by Victor Ruhle (Microsoft Research) 15:33-15:43 Synthetic data by Carmela Troncoso (École polytechnique fédérale de Lausanne) 15:43-15:56 Privacy auditing for ML + federated learning by Reza Shokri (National University of Singapore) 15:56-16:10 Q&A on the technical presentations 16:10-16:15 Break 16:15-17:00 – Roundtable discussion on how PPML techniques should be addressed in ongoing regulatory discussions such as the AI Act, European Health Data Space regulation, moderated by Bianca-Ioana Marcu (Co-director of the Privacy Salon & Legal researcher at VUB) A reception will immediately follow the roundtable. We will do our best to accommodate as many in-person guests as possible, but please note that only a limited number of in-person places are available. In Cooperation with Brussels Privacy Hub of Vrije Universiteit Brussel (VUB) and Microsoft.  |