Efficient Secure CNN Inference: A Multi-Server Framework Based on Conditional Separable and Homomorphic Encryption

被引:0
作者
Sun, Longlong [1 ]
Li, Hui [2 ,3 ]
Peng, Yanguo [2 ]
Cui, Jiangtao [2 ]
机构
[1] Xidian Univ, Sch Cyber Engn, Xian 710071, Shaanxi, Peoples R China
[2] Xidian Univ, Sch Comp Sci & Technol, Xian 710071, Shaanxi, Peoples R China
[3] Yunxi Technol Co Ltd, Guangzhou, Peoples R China
基金
中国国家自然科学基金;
关键词
Servers; Homomorphic encryption; Cloud computing; Protocols; Privacy; Cryptography; Convolutional neural networks; Privacy preservation; convolutional neural networks; homomorphic encryption; secure multi-party computation;
D O I
10.1109/TCC.2024.3443405
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
Deep learning inference has become a fundamental component of cloud service providers, while privacy issues during services have received significant attention. Although many privacy-preserving schemes have been proposed, they require further improvement. In this article, we propose Serpens, an efficient convolutional neural network (CNN) secure inference framework to protect users' uploaded data. We introduce a pair of novel concepts, namely separable and conditional separable, to determine whether a layer in CNNs can be computed over multiple servers or not. We demonstrate that linear layers are separable and construct factor-functions to reduce their overhead to nearly zero. For the two nonlinear layers, i.e., ReLU and max pooling, we design four secure protocols based on homomorphic encryption and random masks for two- and n-server settings. These protocols are essentially different from existing schemes, which are primarily based on garbled circuits. In addition, we extensively propose a method to split the image securely. The experimental results demonstrate that Serpens is 60x-197x faster than the previous scheme in the two-server setting. The superiority of Serpens is even more significant in the n-server setting, only less than an order of magnitude slower than performing plaintext inference over clouds.
引用
收藏
页码:1116 / 1130
页数:15
相关论文
共 46 条
  • [31] An Enhanced and Provably Secure Chaotic Map-Based Authenticated Key Agreement in Multi-Server Architecture
    Irshad, Azeem
    Chaudhry, Shehzad Ashraf
    Xie, Qi
    Li, Xiong
    Farash, Mohammad Sabzinejad
    Kumari, Saru
    Wu, Fan
    ARABIAN JOURNAL FOR SCIENCE AND ENGINEERING, 2018, 43 (02) : 811 - 828
  • [32] Efficient and HRA Secure Universal Conditional Proxy Re-Encryption for Cloud-Based Data Sharing
    Hu, Huidan
    Zhou, Yuanjian
    Cao, Zhenfu
    Dong, Xiaolei
    APPLIED SCIENCES-BASEL, 2022, 12 (19):
  • [33] An Enhanced and Provably Secure Chaotic Map-Based Authenticated Key Agreement in Multi-Server Architecture
    Azeem Irshad
    Shehzad Ashraf Chaudhry
    Qi Xie
    Xiong Li
    Mohammad Sabzinejad Farash
    Saru Kumari
    Fan Wu
    Arabian Journal for Science and Engineering, 2018, 43 : 811 - 828
  • [34] A design of provably secure multi-factor ECC-based authentication protocol in multi-server cloud architecture
    Shivangi Shukla
    Sankita J. Patel
    Cluster Computing, 2024, 27 : 1559 - 1580
  • [35] A design of provably secure multi-factor ECC-based authentication protocol in multi-server cloud architecture
    Shukla, Shivangi
    Patel, Sankita J.
    CLUSTER COMPUTING-THE JOURNAL OF NETWORKS SOFTWARE TOOLS AND APPLICATIONS, 2024, 27 (02): : 1559 - 1580
  • [36] ITSSAKA-MS: An Improved Three-Factor Symmetric-Key Based Secure AKA Scheme for Multi-Server Environments
    Ali, Zeeshan
    Hussain, Sajid
    Rehman, Rana Haseeb Ur
    Munshi, Asmaa
    Liaqat, Misbah
    Kumar, Neeraj
    Chaudhry, Shehzad Ashraf
    IEEE ACCESS, 2020, 8 (08): : 107993 - 108003
  • [37] Secure and Flexible Privacy-Preserving Federated Learning Based on Multi-Key Fully Homomorphic Encryption
    Shen, Jiachen
    Zhao, Yekang
    Huang, Shitao
    Ren, Yongjun
    ELECTRONICS, 2024, 13 (22)
  • [38] Secure and efficient proxy re-encryption scheme based on key-homomorphic constrained PRFs in cloud computing
    Luo, Wei
    Ma, Wenping
    CLUSTER COMPUTING-THE JOURNAL OF NETWORKS SOFTWARE TOOLS AND APPLICATIONS, 2019, 22 (02): : 541 - 551
  • [39] Secure and efficient proxy re-encryption scheme based on key-homomorphic constrained PRFs in cloud computing
    Wei Luo
    Wenping Ma
    Cluster Computing, 2019, 22 : 541 - 551
  • [40] Depth-Wise Separable Convolutions and Multi-Level Pooling for an Efficient Spatial CNN-Based Steganalysis
    Zhang, Ru
    Zhu, Feng
    Liu, Jianyi
    Liu, Gongshen
    IEEE TRANSACTIONS ON INFORMATION FORENSICS AND SECURITY, 2020, 15 : 1138 - 1150