Adaptive Laplace Mechanism: Differential Privacy Preservation in Deep Learning

被引:114
|
作者
Phan, NhatHai [1 ]
Wu, Xintao [2 ]
Hu, Han [1 ]
Dou, Dejing [3 ]
机构
[1] New Jersey Inst Technol, Newark, NJ 07102 USA
[2] Univ Arkansas, Fayetteville, AR 72701 USA
[3] Univ Oregon, Eugene, OR 97403 USA
关键词
D O I
10.1109/ICDM.2017.48
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
In this paper, we focus on developing a novel mechanism to preserve differential privacy in deep neural networks, such that: (1) The privacy budget consumption is totally independent of the number of training steps; (2) It has the ability to adaptively inject noise into features based on the contribution of each to the output; and (3) It could be applied in a variety of different deep neural networks. To achieve this, we figure out a way to perturb affine transformations of neurons, and loss functions used in deep neural networks. In addition, our mechanism intentionally adds "more noise" into features which are "less relevant" to the model output, and vice-versa. Our theoretical analysis further derives the sensitivities and error bounds of our mechanism. Rigorous experiments conducted on MNIST and CIFAR-10 datasets show that our mechanism is highly effective and outperforms existing solutions.
引用
收藏
页码:385 / 394
页数:10
相关论文
共 50 条
  • [31] A hybrid deep learning framework for privacy preservation in edge computing
    Rajashree, R. Harine
    Sundarakantham, K.
    Sivasankar, E.
    Shalinie, S. Mercy
    COMPUTERS & SECURITY, 2023, 129
  • [32] A federated deep learning framework for privacy preservation and communication efficiency
    Cao, Tien-Dung
    Tram, Truong-Huu
    Tran, Hien
    Tran, Khanh
    JOURNAL OF SYSTEMS ARCHITECTURE, 2022, 124
  • [33] The optimal upper bound of the number of queries for Laplace mechanism under differential privacy
    Li, Xiaoguang
    Li, Hui
    Zhu, Hui
    Huang, Muyang
    INFORMATION SCIENCES, 2019, 503 : 219 - 237
  • [34] Optimal Balance of Privacy and Utility with Differential Privacy Deep Learning Frameworks
    Kotevska, Olivera
    Alamudun, Folami
    Stanley, Christopher
    2021 INTERNATIONAL CONFERENCE ON COMPUTATIONAL SCIENCE AND COMPUTATIONAL INTELLIGENCE (CSCI 2021), 2021, : 425 - 430
  • [35] The Laplace Mechanism has optimal utility for differential privacy over continuous queries
    Fernandes, Natasha
    McIver, Annabelle
    Morgan, Carroll
    2021 36TH ANNUAL ACM/IEEE SYMPOSIUM ON LOGIC IN COMPUTER SCIENCE (LICS), 2021,
  • [36] When Deep Learning Meets Differential Privacy: Privacy,Security, and More
    Li, Xinyan
    Chen, Yufei
    Wang, Cong
    Shen, Chao
    IEEE NETWORK, 2021, 35 (06): : 148 - 155
  • [37] Dynamic Personalized Federated Learning with Adaptive Differential Privacy
    Yang, Xiyuan
    Huang, Wenke
    Ye, Mang
    ADVANCES IN NEURAL INFORMATION PROCESSING SYSTEMS 36 (NEURIPS 2023), 2023,
  • [38] An adaptive federated learning scheme with differential privacy preserving
    Wu, Xiang
    Zhang, Yongting
    Shi, Minyu
    Li, Pei
    Li, Ruirui
    Xiong, Neal N.
    FUTURE GENERATION COMPUTER SYSTEMS-THE INTERNATIONAL JOURNAL OF ESCIENCE, 2022, 127 : 362 - 372
  • [39] Bidirectional adaptive differential privacy federated learning scheme
    Li, Yang
    Xu, Jin
    Zhu, Jianming
    Wang, Youwei
    Xi'an Dianzi Keji Daxue Xuebao/Journal of Xidian University, 2024, 51 (03): : 158 - 169
  • [40] Enhancing the Trajectory Privacy with Laplace Mechanism
    Quan, Daiyong
    Yin, Lihua
    Guo, Yunchun
    2015 IEEE TRUSTCOM/BIGDATASE/ISPA, VOL 1, 2015, : 1218 - 1223