Secure Computation for Machine Learning With SPDZ

TL;DR

This paper evaluates the SPDZ framework's efficiency and security in performing privacy-preserving machine learning algorithms like linear and logistic regression, demonstrating improved performance over prior semi-honest MPC methods.

Contribution

It provides an empirical analysis of SPDZ's performance and security in real-world ML tasks, highlighting its scalability and robustness.

Findings

SPDZ outperforms previous semi-honest MPC implementations.

SPDZ offers stronger security guarantees.

Efficient computation of ML algorithms with SPDZ is feasible.

Abstract

Secure Multi-Party Computation (MPC) is an area of cryptography that enables computation on sensitive data from multiple sources while maintaining privacy guarantees. However, theoretical MPC protocols often do not scale efficiently to real-world data. This project investigates the efficiency of the SPDZ framework, which provides an implementation of an MPC protocol with malicious security, in the context of popular machine learning (ML) algorithms. In particular, we chose applications such as linear regression and logistic regression, which have been implemented and evaluated using semi-honest MPC techniques. We demonstrate that the SPDZ framework outperforms these previous implementations while providing stronger security.