analiza aiqbench
This research introduces a masking-based secure aggregation framework designed to mitigate privacy leakage in federated learning models trained on sensitive clinical EEG data. By integrating threshold secret sharing and dropout-resilient mechanisms, the approach effectively hides individual model updates from central servers while maintaining compatibility with standard training workflows, albeit at increased computational overhead for malicious threat scenarios. The study provides a critical advancement for deploying distributed machine learning systems in healthcare environments where data sovereignty and patient privacy are paramount constraints.
Secure Aggregation for Privacy-Preserving Federated Learning on Clinical EEG Data
arXiv:2607.28191v1 Announce Type: cross Abstract: Federated learning enables multiple institutions to train shared models without exchanging raw clinical EEG data, but it does not fully prevent privacy leakage from individual model updates. This paper presents a privacy-preserving federated learning framework for clinical EEG data using masking-based secure aggregation as the core protection mechanism. The framework combines graph-based communication, threshold secret sharing, dropout-resilient aggregation, local update clipping, an optional Bloom filter-based privacy-preserving record-linkage initialization module, and auxiliary-notary-based verifiability. It supports both semi-honest and malicious aggregation settings and is implemented using the Flower federated learning framework. The secure-aggregation variants are evaluated in a simulated cross-silo healthcare setting using TUH EEG-derived data under different client configurations. Under the stated assumptions, the secure variants hide individual updates from the aggregation server. The results show that these variants remain compatible with federated model training, although malicious-setting safeguards and lightweight consistency-checking mechanisms introduce additional computation, communication, and round-duration overhead. The semi-honest variant provides the lowest overhead among the secure configurations, while malicious and auxiliary-notary variants offer stronger consistency, integrity, and lightweight verification support at higher cost.
Abstract:Federated learning enables multiple institutions to train shared models without exchanging raw clinical EEG data, but it does not fully prevent privacy leakage from individual model updates. This paper presents a privacy-preserving federated learning framework for clinical EEG data using masking-based secure aggregation as the core protection mechanism. The framework combines graph-based communication, threshold secret sharing, dropout-resilient aggregation, local update clipping, an optional Bloom filter-based privacy-preserving record-linkage initialization module, and auxiliary-notary-based verifiability. It supports both semi-honest and malicious aggregation settings and is implemented using the Flower federated learning framework. The secure-aggregation variants are evaluated in a simulated cross-silo healthcare setting using TUH EEG-derived data under different client configurations. Under the stated assumptions, the secure variants hide individual updates from the aggregation server. The results show that these variants remain compatible with federated model training, although malicious-setting safeguards and lightweight consistency-checking mechanisms introduce additional computation, communication, and round-duration overhead. The semi-honest variant provides the lowest overhead among the secure configurations, while malicious and auxiliary-notary variants offer stronger consistency, integrity, and lightweight verification support at higher cost.
27 pages, 6 figures, 7 tables. A version of this manuscript has been accepted for presentation at the International Workshop on Hot Topics at the Intersection of Distributed Machine Learning and Security (HotDiSec 2026), co-located with ESORICS 2026
Cryptography and Security (cs.CR); Distributed, Parallel, and Cluster Computing (cs.DC); Machine Learning (cs.LG)
arXiv:2607.28191 [cs.CR]
(or arXiv:2607.28191v1 [cs.CR] for this version)
https://doi.org/10.48550/arXiv.2607.28191 arXiv-issued DOI via DataCite (pending registration)
Text integral sub licența deschisă a publicației · 241 cuvinte
Subiecte conexe
Trustworthy Privacy-Preserving Multimodal Federated Learning for Personalised Breast Cancer Prediction
arXiv · 23 iul. 2026
Biomedical Imaging's Real Bottleneck Is the Data, Not the Model
Databricks · 8 oct. 2026
Aegis: Generative Gradient Masking for Privacy-Preserving Medical Federated Learning
arXiv · 1 oct. 2026
Recovering Clinical Utility Under Differential Privacy: Empirical Validation of Adaptive Federated Aggregation on Heterogeneous Cardiovascular Datasets
arXiv · 23 iul. 2026