Summary of Centaur: Bridging the Impossible Trinity Of Privacy, Efficiency, and Performance in Privacy-preserving Transformer Inference, by Jinglong Luo et al.
Centaur: Bridging the Impossible Trinity of Privacy, Efficiency, and Performance in Privacy-Preserving Transformer Inference
by Jinglong Luo, Guanzhong Chen, Yehong Zhang, Shiyu Liu, Hui Wang, Yue Yu, Xun Zhou, Yuan Qi, Zenglin Xu
First submitted to arxiv on: 14 Dec 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Cryptography and Security (cs.CR)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The proposed Centaur framework addresses the “impossible trinity” of privacy, efficiency, and performance in pre-trained Transformer models deployed for inference services on cloud platforms. By protecting model parameters with random permutations and inference data with Secure Multi-Party Computation (SMPC), Centaur achieves a better balance between these competing factors. The framework leverages the strengths of both techniques to provide strong privacy guarantees while maintaining accurate results and improving inference speed by 5.0-30.4 times compared to plaintext inference. Experimental results demonstrate that Centaur’s privacy protection capabilities can withstand various existing model inversion attack methods. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Pre-trained Transformer models are being used for inference services on cloud platforms, but this raises concerns about the privacy of model parameters and inference data. Researchers have proposed solutions like Secure Multi-Party Computation (SMPC) and random permutations to protect privacy, but these approaches come with trade-offs in terms of efficiency and performance. A new framework called Centaur aims to address these issues by combining the strengths of both techniques. This framework is designed to balance privacy, efficiency, and performance while providing strong protection against model inversion attacks. |
Keywords
» Artificial intelligence » Inference » Transformer
