Computer Science News
UCL-EPFL Team Wins NDSS Distinguished Paper Award
We are very pleased to announce that Apostolos Pyrgelis and Dr Emiliano De Cristofaro, of UCL Department of Computer Science, and their collaborator Dr Carmela Troncoso, of EPFL, have won one of the two Distinguished Paper Awards at the Network and Distributed System Security Symposium (NDSS) 2018.
The NDSS symposium takes place annually in San Diego, and provides a platform for knowledge exchange among researchers and practitioners of network and distributed system security. It is the Internet Society’s flagship security conference, and is lauded as one of the top four academic conferences in security.
The winning paper, entitled “Knock Knock, Who’s There? Membership Inference on Aggregate Location Data”, presented the first study on the feasibility of membership inference attacks on aggregate location time-series.
“It is a real honour to have won a distinguished paper award this year, as there were so many fantastic submissions, said Dr De Cristofaro. “I am very proud of Apostolos’ work on this research, as it demonstrates a very rare and vital combination of foundational insights and real-world work.”
Data aggregation is often considered as a privacy defence mechanism in the context of mobility analytics. However, the team’s paper confirms it is not, as an adversary that can re-identify users whose data is part of the aggregates. In order to show this, they build a framework that includes prior knowledge of the adversary, and that models the attack as an interactive game, which the adversary wins using a machine learning classifier. This framework is applied to differential privacy defences used for time-series release, such as output perturbation, as well as input perturbation, as implemented by companies like Google and Apple.
“Overall, our work means that we now have a way to carefully test the real-world privacy guarantees provided by differential privacy and other defences, as well as the resulting utility.”
Dr Emiliano De Cristofaro is a Reader (Associate Professor) in Security and Privacy Enhancing Technologies at University College London. He conducts research in the field of security and privacy enhancing technologies, more specifically to understand and counter security issues via measurement studies and data-driven analysis, and to tackle problems at the intersection of machine learning and security.
Apostolos Pyrgelis is a PhD candidate based in the Information Security Research Group at UCL. He is co-supervised by Dr De Cristofaro and Dr Gordon Ross. His research interests include applied cryptography, privacy-enhancing technologies, distributed systems as well as privacy-friendly analytics and machine learning applications to security.
Please join us in congratulating them on this fantastic achievement!
You can find an open access copy of the paper on the NDSS Symposium website here.