《计算机应用》唯一官方网站 ›› 2025, Vol. 45 ›› Issue (2): 506-517.DOI: 10.11772/j.issn.1001-9081.2024020249

• 网络空间安全 • 上一篇    

面向遥感数据的基于本地差分隐私的联邦学习隐私保护方案

陈海田1,2,3, 陈学斌1,2,3(), 马锐奎1,2, 张帅华1,2,3   

  1. 1.华北理工大学 理学院,河北 唐山 063210
    2.河北省数据科学与应用重点实验室(华北理工大学),河北 唐山 063210
    3.唐山市数据科学重点实验室(华北理工大学),河北 唐山 063210
  • 收稿日期:2024-03-11 修回日期:2024-04-03 接受日期:2024-04-09 发布日期:2024-06-04 出版日期:2025-02-10
  • 通讯作者: 陈学斌
  • 作者简介:陈海田(1998—),男,湖南娄底人,硕士研究生,CCF会员,主要研究方向:数据安全、隐私保护
    马锐奎(1995—),男,安徽阜阳人,硕士研究生,CCF会员,主要研究方向:数据安全、网络安全
    张帅华(1999—),男,河北石家庄人,硕士研究生,CCF会员,主要研究方向:数据安全、网络安全、隐私保护。
  • 基金资助:
    国家自然科学基金资助项目(U20A20179)

Federated learning privacy protection scheme based on local differential privacy for remote sensing data

Haitian CHEN1,2,3, Xuebin CHEN1,2,3(), Ruikui MA1,2, Shuaihua ZHANG1,2,3   

  1. 1.College of Science,North China University of Science and Technology,Tangshan Hebei 063210,China
    2.Hebei Provincial Key Laboratory of Data Science and Application (North China University of Science and Technology),Tangshan Hebei 063210,China
    3.Tangshan Key Laboratory of Data Science(North China University of Science and Technology),Tangshan Hebei 063210,China
  • Received:2024-03-11 Revised:2024-04-03 Accepted:2024-04-09 Online:2024-06-04 Published:2025-02-10
  • Contact: Xuebin CHEN
  • About author:CHEN Haitian, born in 1998, M. S. candidate. His research interests include data security, privacy protection.
    MA Ruikui, born in 1995, M. S. candidate. His research interests include data security, network security.
    ZHANG Shuaihua, born in 1999, M. S. candidate. His research interests include data security, network security, privacy protection.
  • Supported by:
    National Natural Science Foundation of China(U20A20179)

摘要:

遥感数据具有高度的时空相关性以及复杂的地物特征,使得这些数据的隐私保护面临挑战。联邦学习作为一种旨在保护参与方数据隐私的分布式学习方法,为应对遥感数据隐私保护面对的挑战提供了有效的解决方案;然而,在联邦学习模型的训练阶段,恶意攻击者可能通过反演推断参与者的隐私信息,进而导致敏感信息的泄露。针对遥感数据在联邦学习训练中存在的隐私泄露问题,提出一种基于本地差分隐私的联邦学习隐私保护方案。首先,对模型进行预训练,计算模型的层重要性,并根据层重要性合理分配隐私预算;然后,通过对模型更新进行裁剪变换,并对裁剪值进行自适应随机扰动,实现本地差分隐私保护;最后,在聚合扰动更新时,采用模型校正以进一步提高模型性能。理论分析和仿真结果表明,所提方案不仅能为各参与方提供合适的差分隐私保护,并有效防止通过反演推断出隐私敏感信息,而且在3个遥感数据集上相较于基于分段机制的扰动方案提升了3.28~3.93个百分点的准确率。可见,所提方案在保证隐私的同时有效保障了模型性能。

关键词: 联邦学习, 差分隐私, 层重要性, 遥感数据, 模型校正

Abstract:

Remote sensing data have high spatio-temporal correlation and complex surface features, which makes the privacy protection of the data challenging. As a distributed learning method with the goal of protecting data privacy of the participants, federated learning provides an effective solution to overcome the challenges faced by remote sensing data privacy protection. However, during the training phase of federated learning models, malicious attackers may infer private information of the participants through inversion, leading to the disclosure of sensitive information. Aiming at the privacy leakage problem of remote sensing data in federated learning training, a federated learning privacy protection scheme based on local differential privacy was proposed. Firstly, the model was pre-trained, the layer importance of the model was calculated, and the privacy budget was allocated reasonably based on the layer importance. Then, local differential privacy protection was achieved by performing a crop transformation on the model update and performing adaptive random disturbance on the crop value. Finally, model correction was employed to further improve the model performance when the aggregated disturbance was updated. Theoretical analysis and simulation results show that the proposed scheme can not only provide appropriate differential privacy protection for each participant and prevent inferring privacy sensitive information through inversion effectively, but also outperform the segmentation mechanism-based disturbance scheme in accuracy on three remote sensing datasets by 3.28 to 3.93 percentage points. It can be seen that the proposed scheme guarantees model performance effectively while ensuring privacy.

Key words: federated learning, differential privacy, layer importance, remote sensing data, model correction

中图分类号: