defense arXiv Dec 17, 2025 · Dec 2025
Mukur Gupta, Niharika Gupta, Saifur Rahman et al. · Columbia University · Vellore Institute of Technology +1 more
Defends FL models against adversarial attacks by synthesizing server-side training data from client model trajectories, enabling adversarial training without client data access
Input Manipulation Attack visionfederated-learning
Deep learning models deployed on edge devices are increasingly used in safety-critical applications. However, their vulnerability to adversarial perturbations poses significant risks, especially in Federated Learning (FL) settings where identical models are distributed across thousands of clients. While adversarial training is a strong defense, it is difficult to apply in FL due to strict client-data privacy constraints and the limited compute available on edge devices. In this work, we introduce TrajSyn, a privacy-preserving framework that enables effective server-side adversarial training by synthesizing a proxy dataset from the trajectories of client model updates, without accessing raw client data. We show that TrajSyn consistently improves adversarial robustness on image classification benchmarks with no extra compute burden on the client device.
cnn federated Columbia University · Vellore Institute of Technology · Deakin University
ML Security Papers