Privacy-Preserving Federated Learning

Aurélien BELLET

Date(s) : 25/03/2022   iCal
14 h 30 min - 15 h 30 min

Federated learning (FL) is a machine learning paradigm where several participants collaboratively train a model while keeping their data decentralized. However, the model parameters or gradients exchanged during the FL training process may leak information about the data. After a brief introduction to FL, I will show how to use the notion of Differential Privacy (DP) to design FL algorithms that provably ensure privacy and confidentiality. In particular, I will present two approaches (one for server-orchestrated FL and one for fully decentralized FL) that nearly match the privacy-utility trade-off of the centralized setting without relying on a trusted curator or complex secure computation primitives.



Retour en haut 

Secured By miniOrange