Secure Collaborative Training and Inference for XGBoost
Secure Collaborative Training and Inference for XGBoost
复制标题
XGBoost 的安全协作训练和推理
DOI：
10.1145/3411501.3419420
复制
发表时间：
2020
期刊：
影响因子：
--
通讯作者：
Wenting Zheng
中科院分区：
文献类型：
--
作者：
Andrew Law;Chester Leung;Rishabh Poddar;Raluca A. Popa;Chenyu Shi;Octavian Sima;Chaofan Yu;Xingmeng Zhang;Wenting Zheng
近年来，事实证明，梯度增强的决策树学习是一种有效的培训模型的方法。此外，多个政党之间的协作学习有可能极大地使所有有关方面受益，但是由于业务，监管和责任问题，组织也遇到了共享敏感数据的障碍。我们提出了Secure XGBoost，这是一种隐私保护系统，可实现XGBoost模型的多方培训和推断。 Secure XGBoost在硬件飞地的帮助下保护各方数据的隐私以及计算的完整性。至关重要的是，安全XGBoost使用新颖的数据构成算法增强了飞地的安全性，这些算法可防止访问通过访问模式泄漏引起的飞地的访问侧通道攻击。
In recent years, gradient boosted decision tree learning has proven to be an effective method of training robust models. Moreover, collaborative learning among multiple parties has the potential to greatly benefit all parties involved, but organizations have also encountered obstacles in sharing sensitive data due to business, regulatory, and liability concerns. We propose Secure XGBoost, a privacy-preserving system that enables multiparty training and inference of XGBoost models. Secure XGBoost protects the privacy of each party's data as well as the integrity of the computation with the help of hardware enclaves. Crucially, Secure XGBoost augments the security of the enclaves using novel data-oblivious algorithms that prevent access side-channel attacks on enclaves induced via access pattern leakage.