Paper 2023/647

Efficient FHE-based Privacy-Enhanced Neural Network for AI-as-a-Service

Kwok-Yan Lam, Nanyang Technological University
Xianhui Lu, Chinese Academy of Sciences
Linru Zhang, Nanyang Technological University
Xiangning Wang, Nanyang Technological University
Huaxiong Wang, Nanyang Technological University
Si Qi Goh, Nanyang Technological University

AI-as-a-Service has emerged as an important trend for supporting the growth of the digital economy. Digital service providers make use of their vast amount of user data to train AI models (such as image recognitions, financial modelling and pandemic modelling etc.) and offer them as a service on the cloud. While there are convincing advantages for using such third-party models, the fact that users need to upload their data to the cloud is bound to raise serious privacy concerns, especially in the face of increasingly stringent privacy regulations and legislations. To promote the adoption of AI-as-a-Service while addressing the privacy issues, we propose a practical approach for constructing privacy-enhanced neural networks by designing an efficient implementation of fully homomorphic encryption. With this approach, an existing neural network can be converted to process FHE-encrypted data and produce encrypted output which are only accessible by the model users, and more importantly, within an operationally acceptable time (e.g. within 1 second for facial recognition in typical border control systems). Experimental results show that in many practical tasks such as facial recognition, text classification and so on, we obtained the state-of-the-art inference accuracy in less than one second on a 16 cores CPU.

Available format(s)
Publication info
Fully homomorphic encryptionPrivacy-enhanced neural networksLook-up table algorithmDigital trust
Contact author(s)
kwokyan lam @ ntu edu sg
luxianhui @ iie ac cn
linru zhang @ ntu edu sg
xiangning wang @ ntu edu sg
hxwang @ ntu edu sg
siqi005 @ e ntu edu sg
2023-05-08: approved
2023-05-08: received
See all versions
Short URL
Creative Commons Attribution-NonCommercial-NoDerivs


      author = {Kwok-Yan Lam and Xianhui Lu and Linru Zhang and Xiangning Wang and Huaxiong Wang and Si Qi Goh},
      title = {Efficient FHE-based Privacy-Enhanced Neural Network for AI-as-a-Service},
      howpublished = {Cryptology ePrint Archive, Paper 2023/647},
      year = {2023},
      note = {\url{}},
      url = {}
Note: In order to protect the privacy of readers, does not use cookies or embedded third party content.