debasishetc/PhysioAttack-FL
收藏资源简介:
PhysioAttack-FL是一个用于物联网医疗应用中联邦学习系统对抗鲁棒性评估的基准数据集。该数据集包含在良性和对抗条件下收集的联邦客户端模型更新的特征级表示。基准数据是通过在异构树莓派边缘设备上使用生理压力监测数据集进行联邦学习实验生成的。数据集支持恶意客户端检测、鲁棒聚合、拜占庭容错和安全联邦学习的研究。数据集来源于公开的生理数据集SWELL-KW和WESAD,但仅包含从联邦模型更新中提取的派生统计和几何特征,不重新分发原始生理信号。
PhysioAttack-FL is a benchmark dataset for adversarial robustness evaluation in Federated Learning (FL) systems for Internet of Medical Things (IoMT) applications. The dataset contains feature-level representations of federated client model updates collected under both benign and adversarial conditions. The benchmark was generated from federated learning experiments conducted on heterogeneous Raspberry Pi edge devices using physiological stress monitoring datasets. The dataset supports research on malicious client detection, robust aggregation, Byzantine resilience, and secure federated learning.




