arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2202.07835cs.CRcs.LG
收藏

SecGNN:隐私保护图神经网络训练和推理云服务

SecGNN: Privacy-Preserving Graph Neural Network Training and Inference as a Cloud Service

Harbin Institute of Technology(哈尔滨工业大学);City University of Hong Kong(香港城市大学)

Songlei Wang, Yifeng Zheng, Xiaohua Jia

摘要:

图被广泛用于对实体之间的复杂关系进行建模。图神经网络作为一种强大的图分析工具,由于其端到端的处理能力,近年来得到了广泛的关注。随着云计算的普及,将复杂且资源密集型的模型训练和推理服务部署在云中,因其显著的优势而越来越受欢迎。然而,GNN训练和推理服务如果部署在云中,将引发对信息丰富和专有图数据(以及由此产生的模型)的严重隐私问题。虽然已经有一些关于安全神经网络训练和推理的工作,但它们都专注于处理图像和文本的卷积神经网络,而不是具有丰富结构信息的复杂图形数据。在本文中,我们设计,实现和评估SecGNN,这是第一个在云中支持隐私保护GNN训练和推理服务的系统。SecGNN是基于对轻量级密码学和机器学习技术的见解的协同作用而构建的。我们深入研究了GNN的训练和推理过程,并设计了一系列相应的安全定制协议来支持整体计算。大量的实验表明,SecGNN实现了相当的明文训练和推理精度,具有良好的性能。

英文摘要:

Graphs are widely used to model the complex relationships among entities. As a powerful tool for graph analytics, graph neural networks (GNNs) have recently gained wide attention due to its end-to-end processing capabilities. With the proliferation of cloud computing, it is increasingly popular to deploy the services of complex and resource-intensive model training and inference in the cloud due to its prominent benefits. However, GNN training and inference services, if deployed in the cloud, will raise critical privacy concerns about the information-rich and proprietary graph data (and the resulting model). While there has been some work on secure neural network training and inference, they all focus on convolutional neural networks handling images and text rather than complex graph data with rich structural information. In this paper, we design, implement, and evaluate SecGNN, the first system supporting privacy-preserving GNN training and inference services in the cloud. SecGNN is built from a synergy of insights on lightweight cryptography and machine learning techniques. We deeply examine the procedure of GNN training and inference, and devise a series of corresponding secure customized protocols to support the holistic computation. Extensive experiments demonstrate that SecGNN achieves comparable plaintext training and inference accuracy, with promising performance.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑