The lab is currently involved in the following projects:
EPFL COVID-19 Real Time Epidemiology I-DAIR Pathfinder
This is an interdisciplinary project funded by the Botnar Foundation in which we aim at developing better technology for digital epidemiology, such as new proximity tracing technologies and protocols, as well as new data mining techniques to predict epidemiological events. Our partners in this project are: HexHive@EPFL (Mathias Payer), VLSC@EPFL (Jim Larus), MLO@EPFL (Martin Jaggi), DCSL@EPFL (Ed Bugnion), Salathe Lab@EPFL (Marcel Salathe), System Security Group@ETHZ (Srdjan Capkun), Michael Veale @UCL, Seda Gurses@TU Delft, EssentialTech@EPFL, and 3dB Access.
PriBAD: Private Biometrics for Aid
This is a close collaboration with the International Committee of the Red Cross (ICRC) Data Protection Office to build respectful biometrics tailored to their use cases in the field that they can deploy without increasing the risks for their beneficiaries
Automated decryption of ACARS communications
This is a collaboration with Armasuisse in which we aim to develop methods to automatically identify the use of weak encryption in aircraft to ground communications. We also work on automated decryption methods based on machine-learning. In this project we process ACARS messages collected in the wild.
VaultML: Preventing privacy leaks in machine learning
This is a fundamental research project funded by the Swiss National Science Foundation. The overall objective of VaultML is to understand and prevent privacy issues at the output of machine learning processes. Concretely, we have two goals: 1) to develop measurement tools for quantifying the privacy of machine learning models; and 2) to develop defenses that, applied on models or on user’s input to the models under deployment mitigate the privacy impact stemming from the pervasive use of machine learning.