کنفرانس مهندسی برق ایران

صفحه اصلی / سی و چهارمین کنفرانس بین المللی مهندسی برق

Collaborative Model Verification in Federated Learning

نویسندگان :

Hasti Rostami¹ Mohammad Mahdi Mojahedian² Mohammad Reza Aref³

1- دانشگاه صنعتی شریف 2- دانشگاه صنعتی شریف 3- دانشگاه صنعتی شریف

کلمات کلیدی :

arithmetic circuit،cooperative local model verification،malicious server،malicious client،zero knowledge proof

چکیده :

Federated learning (FL) enables clients to collaboratively train a shared machine learning model while preserving data privacy through secure aggregation of local updates. However, traditional approaches that rely solely on the server for verifying the integrity of these updates suffer from significant limitations, including reliance on a single point of failure and susceptibility to centralized attacks. To address these challenges, we propose a framework that distributes the local model verification process among all participants in the FL system. In this framework, clients and the server collaboratively validate updates using secret-shared non-interactive proofs (SNIPs). This decentralized approach ensures robustness against malicious central server and clients, as no single entity has complete control over the verification process.