Federated Learning Differential Privacy Preservation Method Based on Differentiated Noise Addition

被引:2
|
作者
Han, Liquan [1 ]
Fan, Di [1 ]
Liu, Jinyuan [1 ]
Du, Wei [1 ]
机构
[1] Changchun Univ Technol, Coll Comp Sci & Engn, Changchun, Peoples R China
关键词
federated learning; differential privacy; privacy protection;
D O I
10.1109/ICCCBDA56900.2023.10154864
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Differential privacy is an essential tool in federated learning privacy preservation. However, existing differential privacy-preserving techniques introduce excessive noise perturbations for privacy-preserving purposes, which makes federated learning models less usable. Under the condition of satisfying differential privacy, this paper proposes a new noise addition method, differential privacy preservation by federated learning based on differentiated noise addition (DDPFL), to reduce the loss of model usability. Specifically, the method derives importance coefficients for each model parameter by analyzing the value of the gradient update size, the magnitude of the absolute value of the weight parameter, and the relationship between the global model and the local model gradient update trend during the training process for each client. Then, noise is added to the model parameters using a differentiated noise addition mechanism according to the magnitude of the importance coefficients to achieve noise perturbation of the local model. Finally, the effectiveness of the proposed method is validated by an experimental comparison of the MNIST dataset and the Fashion-Mnist dataset. The usability of the model improves under the same differential privacy condition as the federated learning differential privacy preserving method before the improvement.
引用
收藏
页码:285 / 289
页数:5
相关论文
共 50 条
  • [1] Efficient federated learning privacy preservation method with heterogeneous differential privacy
    Ling, Jie
    Zheng, Junchang
    Chen, Jiahui
    [J]. COMPUTERS & SECURITY, 2024, 139
  • [3] A Differential Privacy Federated Learning Scheme Based on Adaptive Gaussian Noise
    Jiao, Sanxiu
    Cai, Lecai
    Wang, Xinjie
    Cheng, Kui
    Gao, Xiang
    [J]. CMES-COMPUTER MODELING IN ENGINEERING & SCIENCES, 2024, 138 (02): : 1679 - 1694
  • [4] Signal Modulation Recognition Method Based on Differential Privacy Federated Learning
    Shi, Jibo
    Qi, Lin
    Li, Kuixian
    Lin, Yun
    [J]. WIRELESS COMMUNICATIONS & MOBILE COMPUTING, 2021, 2021
  • [5] A Personalized Federated Learning Method Based on Knowledge Distillation and Differential Privacy
    Jiang, Yingrui
    Zhao, Xuejian
    Li, Hao
    Xue, Yu
    [J]. ELECTRONICS, 2024, 13 (17)
  • [6] Kalman Filter-Based Differential Privacy Federated Learning Method
    Yang, Xiaohui
    Dong, Zijian
    [J]. APPLIED SCIENCES-BASEL, 2022, 12 (15):
  • [7] A Differential Privacy Federated Learning Scheme with Improved Noise Perturbation
    Liu, Chang
    He, Xiaowei
    Wang, Bin
    Sun, Xinru
    Luo, Yixuan
    Zeng, Yiji
    Wang, Xinyu
    Wang, Jianhang
    Zhao, Haofei
    [J]. ADVANCED INTELLIGENT COMPUTING TECHNOLOGY AND APPLICATIONS, PT IX, ICIC 2024, 2024, 14870 : 61 - 71
  • [8] Privacy Preservation Method for Vertical Federated Learning Based on Max-min Strategy
    Li, Rong-Chang
    Liu, Tao
    Zheng, Hai-Bin
    Chen, Jin-Yin
    Liu, Zhen-Guang
    Ji, Shou-Ling
    [J]. Zidonghua Xuebao/Acta Automatica Sinica, 2024, 50 (07): : 1373 - 1388
  • [9] Group Signature Based Federated Learning Approach for Privacy Preservation
    Kanchan, Sneha
    Choi, Bong Jun
    [J]. INTERNATIONAL CONFERENCE ON ELECTRICAL, COMPUTER AND ENERGY TECHNOLOGIES (ICECET 2021), 2021, : 1882 - 1887
  • [10] Privacy preservation-based federated learning with uncertain data
    Cao, Fan
    Liu, Bo
    He, Jinghui
    Xu, Jian
    Xiao, Yanshan
    [J]. INFORMATION SCIENCES, 2024, 678