CIS - "Get to know your neighbors" Seminar series - Prof. Jean-Pierre Hubaux

 
Title: Secure and Privacy-Conscious Federated Analytics
Jean-Pierre Hubaux, Head of Laboratory for Data Security (LDS), EPFL

Abstract: In this talk, we address the problem of privacy-preserving training and evaluation of neural networks in an N-party, federated learning setting. We propose a novel system, POSEIDON, the first of its kind in the regime of privacy-preserving neural network training. It employs multiparty lattice-based cryptography to preserve the confidentiality of the training data, the model, and the evaluation data, under a passive-adversary model and collusions between up to N−1 parties. It relies on homomorphic encryption and secure multi-party computation. We also introduce arbitrary linear transformations within the cryptographic bootstrapping operation, optimizing the costly cryptographic computations over the parties. We also mention Lattigo, our quantum-resistant open-source cryptographic library on which POSEIDON is based. Our experimental results show that POSEIDON achieves accuracy similar to centralized (or decentralized) non-private approaches and that its computation and communication overhead scales linearly with the number of parties. Furthermore, we explain how we are using these techniques for the federated analysis of medical data, in particular for genome-wide association studies. Finally, we mention our joint work with lawyers showing GDPR compliance,  and address the creation of start-up Tune Insight.

Bio: Jean-Pierre Hubaux is a full professor at EPFL and head of the Laboratory for Data Security. Through his research, he contributes to laying the foundations and developing the tools for protecting privacy in today’s hyper-connected world. He has pioneered the areas of privacy and security in mobile/wireless networks and in personalized health.
He is the academic director of the Center for Digital Trust (C4DT). He leads the  Data Protection in Personalized Health (DPPH) project funded by the ETH Council. He is a Fellow of both IEEE (2008) and ACM (2010). Recent awards: three of his papers obtained distinctions at the IEEE Symposium on Security and Privacy in 2015, 2018 and 2021. He is among the most cited researchers in privacy protection and in information security. More about him here.

 
The CIS seminar will take place live on Zoom: https://epfl.zoom.us/j/65412917575

Please connect to your zoom account using your "@epfl.ch" address, as this live event is only open to the EPFL community
Monday, October 11th, 2021 from 3:15 to 4:15 pm
NB: Video recordings of the seminars will be made available on our website and published on our social media pages