Privacy Assessment of Federated Learning using Private Personalized Layers

Théo Jourdan (1, 2) , Antoine Boutet (1) , Carole Frindel (3, 2)

1 PRIVATICS - Privacy Models, Architectures and Tools for the Information Society
2 MYRIAD - Modeling & analysis for medical imaging and Diagnosis
3 CREATIS - Centre de Recherche en Acquisition et Traitement de l'Image pour la Santé

Federated Learning (FL) is a collaborative scheme to train a learning model across multiple participants without sharing data. While FL is a clear step forward towards enforcing usersâĂŹ privacy, different inference attacks have been developed. In this paper, we quantify the utility and privacy trade-off of a FL scheme using private personalized layers. While this scheme has been proposed as local adaptation to improve the accuracy of the model through local personalization, it has also the advantage to minimize the information about the model exchanged with the server. However, the privacy of such a scheme has never been quantified. Our evaluations using motion sensor dataset show that personalized layers speed up the convergence of the model and slightly improve the accuracy for all users compared to a standard FL scheme while better preventing both attribute and membership inferences compared to a FL scheme using local differential privacy.

Mots clés: Privacy; Federated Learning; Inference Attacks

Domaines: Traitement du signal et de l'image [eess.SP]; Cryptographie et sécurité [cs.CR]

Théo Jourdan, Antoine Boutet, Carole Frindel. Privacy Assessment of Federated Learning using Private Personalized Layers. MLSP 2021 - IEEE International Workshop on Machine Learning for Signal Processing, Oct 2021, Queensland, Australia. pp.1-5, ⟨10.1109/MLSP52302.2021.9596237⟩. ⟨hal-03354722⟩