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 Tools annual

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 Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Recently, Graph Neural Networks (GNNs), including Homogeneous Graph NeuralNetworks (HomoGNNs) and Heterogeneous Graph Neural Networks (HeteGNNs), havemade remarkable progress in many physical scenarios, especially incommunication applications. Despite achieving great success, the privacy issueof such models has also received considerable attention. Previous studies haveshown that given a well-fitted target GNN, the attacker can reconstruct thesensitive training graph of this model via model inversion attacks, leading tosignificant privacy worries for the AI service provider. We advocate that thevulnerability comes from the target GNN itself and the prior knowledge aboutthe shared properties in real-world graphs. Inspired by this, we propose anovel model inversion attack method on HomoGNNs and HeteGNNs, namely HomoGMIand HeteGMI. Specifically, HomoGMI and HeteGMI are gradient-descent-basedoptimization methods that aim to maximize the cross-entropy loss on the targetGNN and the $1^{st}$ and $2^{nd}$-order proximities on the reconstructed graph.Notably, to the best of our knowledge, HeteGMI is the first attempt to performmodel inversion attacks on HeteGNNs. Extensive experiments on multiplebenchmarks demonstrate that the proposed method can achieve better performancethan the competitors.

Model Inversion Attacks on Homogeneous and Heterogeneous Graph Neural Networks