DIGITAL.CSIC. Repositorio Institucional del CSIC
Publication
Artículo científico (JournalArticle)
Metric-privacy-inspired noise calibration in federated learning: Improving convergence and preventing client inference attacks
Federated learning (FL) enables the training of a global model across multiple data owners (clients) without sharing raw data. This distributed architecture is orchestrated by a central server that aggregates the local models from the clients. In cases where the server is trusted but not all network nodes, differential privacy (DP) can be used to privatize the aggregated model by adding noise. However, this may affect convergence across the FL rounds. In this work, we build on the notion of metric-privacy as a design principle to calibrate the noise added by the server under a global-DP setting, with the objective of mitigating its impact on the convergence of the aggregated model. We do not enforce metric-privacy as a formal guarantee, but rather use it to guide noise calibration. We compare our approach with vanilla FL and global-DP by analyzing the impact on six aggregation strategies and applying it to a medical imaging use case, simulating different scenarios with homogeneous and non-i.i.d. clients. Finally, we introduce the client inference attack (CIA), where a semi-honest client tries to find whether another client participated in the training and study how it can be mitigated using DP and metric-aware noise calibration. Our experiments show that metric-privacy aware noise calibration strategy improves the accuracy compared to standard DP in all the scenarios analyzed, while achieving a comparable success rate against CIA. These results indicate that metric-privacy inspired noise calibration can deliver a superior utility-privacy trade-off in medical-imaging federated settings., Judith Sáinz-Pardo Díaz and Álvaro López García would like to thank the funding through the project AI4EOSC “Artificial Intelligence for the European Open Science Cloud” that has received funding from the European Union’s Horizon Europe research and innovation programme under grant agreement number 101058593 and from the SIESTA project “Secure Interactive Environments for Sensitive daTa Analytics”, funded by the European Union (Horizon Europe) under grant agreement number 101131957. The work of Andreas Athanasiou is supported by the project RECITALS, funded by the EU (grant agreement 101168490) and the project CRYPTECS, funded by the ANR (project number ANR-20-CYAL-0006) and by the BMBF (project number 16KIS1439). The work of Kangsoo Jung is supported by the project ELSA, funded by the Horizon Europe Framework (with project number 101070617). The work of Catuscia Palamidessi is supported by the project HYPATIA, funded by the ERC (grant agreement number 835294)., Peer reviewed