QUOTIENT: Two-Party Secure Neural Network Training and Prediction

TL;DR

QUOTIENT introduces a novel secure two-party training protocol for discretized deep neural networks, significantly enhancing training efficiency and accuracy in privacy-preserving machine learning.

Contribution

The paper presents QUOTIENT, a new discretized training method combined with a secure protocol, optimizing secure two-party DNN training with state-of-the-art techniques.

Findings

50X reduction in WAN training time

6% increase in model accuracy

Effective integration of normalization and adaptive gradients

Abstract

Recently, there has been a wealth of effort devoted to the design of secure protocols for machine learning tasks. Much of this is aimed at enabling secure prediction from highly-accurate Deep Neural Networks (DNNs). However, as DNNs are trained on data, a key question is how such models can be also trained securely. The few prior works on secure DNN training have focused either on designing custom protocols for existing training algorithms, or on developing tailored training algorithms and then applying generic secure protocols. In this work, we investigate the advantages of designing training algorithms alongside a novel secure protocol, incorporating optimizations on both fronts. We present QUOTIENT, a new method for discretized training of DNNs, along with a customized secure two-party protocol for it. QUOTIENT incorporates key components of state-of-the-art DNN training such as layer normalization and adaptive gradient methods, and improves upon the state-of-the-art in DNN training in two-party computation. Compared to prior work, we obtain an improvement of 50X in WAN time and 6% in absolute accuracy.