Paper 2022/207
Cheetah: Lean and Fast Secure Two-Party Deep Neural Network Inference
Zhicong Huang, Wen-jie Lu, Cheng Hong, and Jiansheng Ding
Abstract
Secure two-party neural network inference (2PC-NN) can offer privacy protection for both the client and the server and is a promising technique in the machine-learning-as-a-service setting. However, the large overhead of the current 2PC-NN in- ference systems is still being a headache, especially when applied to deep neural networks such as ResNet50. In this work, we present Cheetah, a new 2PC-NN inference system that is faster and more communication-efficient than state-of-the-arts. The main contributions of Cheetah are two-fold: the first part includes carefully designed homomorphic encryption-based protocols that can evaluate the linear layers (namely convolution, batch normalization, and fully-connection) without any expensive rotation operation. The second part includes several lean and communication-efficient primitives for the non-linear functions (e.g., ReLU and truncation). Using Cheetah, we present intensive benchmarks over several large-scale deep neural networks. Take ResNet50 for an example, an end- to-end execution of Cheetah under a WAN setting costs less than 2.5 minutes and 2.3 gigabytes of communication, which outperforms CrypTFlow2 (ACM CCS 2020) by about 5.6× and 12.9×, respectively.
Metadata
- Available format(s)
-
PDF
- Category
- Applications
- Publication info
- Published elsewhere. Minor revision. USENIX Security'22
- Keywords
- secure neural inferencesecure two-party computationprivacy-preserving machine learning
- Contact author(s)
- juhou lwj @ alibaba-inc com
- History
- 2022-03-24: last of 5 revisions
- 2022-02-20: received
- See all versions
- Short URL
- https://ia.cr/2022/207
- License
-
CC BY
BibTeX
@misc{cryptoeprint:2022/207,
author = {Zhicong Huang and Wen-jie Lu and Cheng Hong and Jiansheng Ding},
title = {Cheetah: Lean and Fast Secure Two-Party Deep Neural Network Inference},
howpublished = {Cryptology ePrint Archive, Paper 2022/207},
year = {2022},
note = {\url{https://eprint.iacr.org/2022/207}},
url = {https://eprint.iacr.org/2022/207}
}
