BART-FL: A Backdoor Attack-Resilient Federated Aggregation Technique for Cross-Silo Applications

BART-FL (Backdoor-Aware Robust Training for Federated Learning) is a novel defense-oriented framework that enhances the robustness of Federated Learning (FL) against backdoor and poisoning attacks. It integrates Principal Component Analysis (PCA) and clustering-based filtering to isolate and suppress malicious client updates while maintaining accuracy on clean data. Designed for cross-device federated environments, BART-FL provides explainable and privacy-aware aggregation mechanisms to improve resilience against adversarial behavior.

Paper: https://ieeexplore.ieee.org/document/11172307
More relvant research: https://www.solidlab.network   

Acknowledgements This work is based upon the work supported by the National Center for Transportation Cybersecurity and Resiliency (TraCR) (a U.S. Department of Transportation National University Transportation Center) headquartered at Clemson University, Clemson, South Carolina, USA. Any opinions, findings, conclusions, and recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of TraCR, and the U.S. Government assumes no liability for the contents or use thereof.