Robust Federated Learning
Robust federated learning (FL) aims to improve the resilience and accuracy of collaborative model training across decentralized devices while preserving data privacy. Current research heavily focuses on mitigating various attacks, including poisoning and gradient leakage, through techniques like secure aggregation, credibility management, and fine-grained learnable masks, often incorporating primal-dual optimization or knowledge distillation methods. These advancements are crucial for enabling trustworthy and reliable FL deployments in sensitive applications like medical diagnostics and network security, where data privacy and model robustness are paramount.
20papers
Papers
February 12, 2025
November 5, 2024
October 23, 2024
September 22, 2023
August 31, 2023
August 21, 2023
November 27, 2022