English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Recently cloud-based graph convolutional network (GCN) has demonstrated greatsuccess and potential in many privacy-sensitive applications such as personalhealthcare and financial systems. Despite its high inference accuracy andperformance on cloud, maintaining data privacy in GCN inference, which is ofparamount importance to these practical applications, remains largelyunexplored. In this paper, we take an initial attempt towards this and develop$\textit{CryptoGCN}$--a homomorphic encryption (HE) based GCN inferenceframework. A key to the success of our approach is to reduce the tremendouscomputational overhead for HE operations, which can be orders of magnitudehigher than its counterparts in the plaintext space. To this end, we develop anapproach that can effectively take advantage of the sparsity of matrixoperations in GCN inference to significantly reduce the computational overhead.Specifically, we propose a novel AMA data formatting method and associatedspatial convolution methods, which can exploit the complex graph structure andperform efficient matrix-matrix multiplication in HE computation and thusgreatly reduce the HE operations. We also develop a co-optimization frameworkthat can explore the trade offs among the accuracy, security level, andcomputational overhead by judicious pruning and polynomial approximation ofactivation module in GCNs. Based on the NTU-XVIEW skeleton joint dataset, i.e.,the largest dataset evaluated homomorphically by far as we are aware of, ourexperimental results demonstrate that $\textit{CryptoGCN}$ outperformsstate-of-the-art solutions in terms of the latency and number of homomorphicoperations, i.e., achieving as much as a 3.10$\times$ speedup on latency andreduces the total Homomorphic Operation Count by 77.4\% with a small accuracyloss of 1-1.5$\%$.

CryptoGCN: Fast and Scalable Homomorphically Encrypted Graph Convolutional Network Inference