APPFLx:提供隐私保护的跨孤岛联邦学习即服务
机器学习
2023-08-21 v1
摘要
跨孤岛隐私保护联邦学习(PPFL)是一种无需共享敏感(如医疗或金融)本地数据即可协作训练鲁棒且泛化机器学习(ML)模型的强大工具。为简化并加速PPFL的采用,我们推出APPFLx,一个提供隐私保护跨孤岛联邦学习即服务的即用平台。APPFLx采用Globus认证,使用户能够轻松且安全地邀请可信协作者进行PPFL,实现了多种同步与异步FL算法,简化了FL实验启动流程,并支持追踪与可视化FL实验生命周期,使领域专家与ML从业者能在一个平台上轻松编排与评估跨孤岛FL。APPFLx可在线访问:https://appflx.link
引用
@article{arxiv.2308.08786,
title = {APPFLx: Providing Privacy-Preserving Cross-Silo Federated Learning as a Service},
author = {Zilinghan Li and Shilan He and Pranshu Chaturvedi and Trung-Hieu Hoang and Minseok Ryu and E. A. Huerta and Volodymyr Kindratenko and Jordan Fuhrman and Maryellen Giger and Ryan Chard and Kibaek Kim and Ravi Madduri},
journal= {arXiv preprint arXiv:2308.08786},
year = {2023}
}