HeFUN: Homomorphic Encryption for Unconstrained Secure Neural Network Inference
Publication Name
Future Internet
Abstract
Homomorphic encryption (HE) has emerged as a pivotal technology for secure neural network inference (SNNI), offering privacy-preserving computations on encrypted data. Despite active developments in this field, HE-based SNNI frameworks are impeded by three inherent limitations. Firstly, they cannot evaluate non-linear functions such as (Formula presented.), the most widely adopted activation function in neural networks. Secondly, the permitted number of homomorphic operations on ciphertexts is bounded, consequently limiting the depth of neural networks that can be evaluated. Thirdly, the computational overhead associated with HE is prohibitively high, particularly for deep neural networks. In this paper, we introduce a novel paradigm designed to address the three limitations of HE-based SNNI. Our approach is an interactive approach that is solely based on HE, called iLHE. Utilizing the idea of iLHE, we present two protocols: (Formula presented.), which facilitates the direct evaluation of the (Formula presented.) function on encrypted data, tackling the first limitation, and (Formula presented.), which extends the feasible depth of neural network computations and mitigates the computational overhead, thereby addressing the second and third limitations. Based on (Formula presented.) and (Formula presented.) protocols, we build a new framework for SNNI, named (Formula presented.). We prove that our protocols and the (Formula presented.) framework are secure in the semi-honest security model. Empirical evaluations demonstrate that (Formula presented.) surpasses current HE-based SNNI frameworks in multiple aspects, including security, accuracy, the number of communication rounds, and inference latency. Specifically, for a convolutional neural network with four layers on the MNIST dataset, (Formula presented.) achieves (Formula presented.) accuracy with an inference latency of (Formula presented.) s, surpassing the popular HE-based framework CryptoNets proposed by Gilad-Bachrach, which achieves (Formula presented.) accuracy with an inference latency of (Formula presented.) s.
Open Access Status
This publication may be available as open access
Volume
15
Issue
12
Article Number
407