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
  • [1] Differential privacy scheme using Laplace mechanism and statistical method computation in deep neural network for privacy preservation
    Kumar, G. Sathish
    Premalatha, K.
    Maheshwari, G. Uma
    Kanna, P. Rajesh
    Vijaya, G.
    Nivaashini, M.
    [J]. ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE, 2024, 128
  • [2] Differential Privacy Preservation in Deep Learning: Challenges, Opportunities and Solutions
    Zhao, Jingwen
    Chen, Yunfang
    Zhang, Wei
    [J]. IEEE ACCESS, 2019, 7 : 48901 - 48911
  • [3] Deep Learning: Differential Privacy Preservation in the Era of Big Data
    Vasa, Jalpesh
    Thakkar, Amit
    [J]. JOURNAL OF COMPUTER INFORMATION SYSTEMS, 2023, 63 (03) : 608 - 631
  • [4] Adaptive Clipping Bound of Deep Learning with Differential Privacy
    Hu, Yuhang
    Li, De
    Tan, Zhou
    Li, Xianxian
    Wang, Jinyan
    [J]. 2021 IEEE 20TH INTERNATIONAL CONFERENCE ON TRUST, SECURITY AND PRIVACY IN COMPUTING AND COMMUNICATIONS (TRUSTCOM 2021), 2021, : 428 - 435
  • [5] Input Validation for the Laplace Differential Privacy Mechanism
    Costea, Sergiu
    Tapus, Nicolae
    [J]. 2015 20TH INTERNATIONAL CONFERENCE ON CONTROL SYSTEMS AND COMPUTER SCIENCE, 2015, : 469 - 474
  • [6] Conducting Correlated Laplace Mechanism for Differential Privacy
    Wang, Hao
    Xu, Zhengquan
    Xiong, Lizhi
    Wang, Tao
    [J]. CLOUD COMPUTING AND SECURITY, PT II, 2017, 10603 : 72 - 85
  • [7] Differential Privacy via a Truncated and Normalized Laplace Mechanism
    Croft, William
    Sack, Jorg-Rudiger
    Shi, Wei
    [J]. JOURNAL OF COMPUTER SCIENCE AND TECHNOLOGY, 2022, 37 (02) : 369 - 388
  • [8] Differential Privacy via a Truncated and Normalized Laplace Mechanism
    William Croft
    Jörg-Rüdiger Sack
    Wei Shi
    [J]. Journal of Computer Science and Technology, 2022, 37 : 369 - 388
  • [9] Improving Laplace Mechanism of Differential Privacy by Personalized Sampling
    Huang, Wen
    Zhou, Shijie
    Zhu, Tianqing
    Liao, Yongjian
    Wu, Chunjiang
    Qiu, Shilin
    [J]. 2020 IEEE 19TH INTERNATIONAL CONFERENCE ON TRUST, SECURITY AND PRIVACY IN COMPUTING AND COMMUNICATIONS (TRUSTCOM 2020), 2020, : 623 - 630
  • [10] Deep Learning with Differential Privacy
    Abadi, Martin
    Chu, Andy
    Goodfellow, Ian
    McMahan, H. Brendan
    Mironov, Ilya
    Talwar, Kunal
    Zhang, Li
    [J]. CCS'16: PROCEEDINGS OF THE 2016 ACM SIGSAC CONFERENCE ON COMPUTER AND COMMUNICATIONS SECURITY, 2016, : 308 - 318