Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

The Communication-Friendly Privacy-Preserving Machine Learning against Malicious Adversaries

Tianpei Lu, Bingsheng Zhang, Lichun Li, Kui Ren

TL;DR

This work addresses the performance overhead of existing maliciously secure protocols, particularly in finite rings like $\mathbb{Z}_{2^\ell}$, by introducing an efficient protocol for secure linear function evaluation and implements the maliciously secure MPC protocol on GPUs, significantly improving its efficiency and scalability.

Abstract

With the increasing emphasis on privacy regulations, such as GDPR, protecting individual privacy and ensuring compliance have become critical concerns for both individuals and organizations. Privacy-preserving machine learning (PPML) is an innovative approach that allows for secure data analysis while safeguarding sensitive information. It enables organizations to extract valuable insights from data without compromising privacy. Secure multi-party computation (MPC) is a key tool in PPML, as it allows multiple parties to jointly compute functions without revealing their private inputs, making it essential in multi-server environments. We address the performance overhead of existing maliciously secure protocols, particularly in finite rings like $\mathbb{Z}_{2^\ell}$, by introducing an efficient protocol for secure linear function evaluation. We implement our maliciously secure MPC protocol on GPUs, significantly improving its efficiency and scalability. We extend the protocol to handle linear and non-linear layers, ensuring compatibility with a wide range of machine-learning models. Finally, we comprehensively evaluate machine learning models by integrating our protocol into the workflow, enabling secure and efficient inference across simple and complex models, such as convolutional neural networks (CNNs).

The Communication-Friendly Privacy-Preserving Machine Learning against Malicious Adversaries

TL;DR

This work addresses the performance overhead of existing maliciously secure protocols, particularly in finite rings like , by introducing an efficient protocol for secure linear function evaluation and implements the maliciously secure MPC protocol on GPUs, significantly improving its efficiency and scalability.

Abstract

With the increasing emphasis on privacy regulations, such as GDPR, protecting individual privacy and ensuring compliance have become critical concerns for both individuals and organizations. Privacy-preserving machine learning (PPML) is an innovative approach that allows for secure data analysis while safeguarding sensitive information. It enables organizations to extract valuable insights from data without compromising privacy. Secure multi-party computation (MPC) is a key tool in PPML, as it allows multiple parties to jointly compute functions without revealing their private inputs, making it essential in multi-server environments. We address the performance overhead of existing maliciously secure protocols, particularly in finite rings like , by introducing an efficient protocol for secure linear function evaluation. We implement our maliciously secure MPC protocol on GPUs, significantly improving its efficiency and scalability. We extend the protocol to handle linear and non-linear layers, ensuring compatibility with a wide range of machine-learning models. Finally, we comprehensively evaluate machine learning models by integrating our protocol into the workflow, enabling secure and efficient inference across simple and complex models, such as convolutional neural networks (CNNs).

Paper Structure

This paper contains 12 sections, 4 theorems, 27 equations, 14 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Preliminaries
  3. 3PC with Malicious Security
  4. Enhancing PPML.
  5. Dealing with linear operation.
  6. Secure Truncation Protocol.
  7. Secure Non-linear Function Evaluation.
  8. Implementation and Benchmarks
  9. Multiplication performance comparison
  10. The inference of neural network.
  11. related work
  12. Conclusion

Key Result

Lemma 1

Suppose protocol $\Pi_\mathsf{Trans}$ take $\{\langle x^{(i)} \rangle,\langle y^{(i)} \rangle, \langle z^{(i)} \rangle\}_{i \in \mathbb{Z}_{|\mathcal{G}|}}$ as input, and it outputs $\{\langle x'^{(i)} \rangle^{\ell[x]},\langle y^{(i)} \rangle^{\ell[x]}\}_{i \in \mathbb{Z}_{|\mathcal{G}|}}; \lang

Figures (14)

  • Figure 1: Compression of Multiplication Triples.
  • Figure 2: The Inner Product Dimension Reduction Protocol
  • Figure 3: The Inner Product Verification Protocol
  • Figure 4: The Batch Multiplication Verification Protocol
  • Figure 5: The Multiplication Protocol
  • ...and 9 more figures

Theorems & Definitions (8)

  • Lemma 1
  • proof
  • Lemma 2
  • proof
  • Lemma 3
  • proof
  • Theorem 1
  • proof