Graph neural network (GNN) with a powerful representation capability has been widely applied to various areas. Recent works have exposed that GNN is vulnerable the backdoor attack, i.e., models trained maliciously crafted training samples are easily fooled by patched samples. Most of proposed studies launch attack using trigger either randomly generated subgraph [e.g., erdős-rényi (ER-B)] for less computational burden or gradient-based generative graph trojaning (GTA)] enable more effective...

Graph neural network (GNN) has achieved great success on graph representation learning. Challenged by large-scale private data collected from user side, GNN may not be able to reflect the excellent performance, without rich features and complete adjacent relationships. Addressing problem, vertical federated learning (VFL) is proposed implement local protection through training a global model collaboratively. Consequently, for graph-structured data, it natural idea construct GNN-based VFL...

Link prediction, inferring the undiscovered or potential links of graph, is widely applied in real world. By facilitating labeled graph as training data, numerous deep learning-based link prediction methods have been studied, which dominant accuracy compared with nondeep methods. However, threats maliciously crafted graphs will leave a specific backdoor model; thus, when some examples are fed into model, it make wrong defined attack. It an important aspect that has overlooked current...

Graph neural network (GNN) has achieved great success on graph representation learning. Challenged by large scale private data collected from user-side, GNN may not be able to reflect the excellent performance, without rich features and complete adjacent relationships. Addressing problem, vertical federated learning (VFL) is proposed implement local protection through training a global model collaboratively. Consequently, for graph-structured data, it natural idea construct based VFL...

Graph neural network (GNN) has captured wide attention due to its capability of graph representation learning for graph-structured data. However, the distributed data silos limit performance GNN. Vertical federated (VFL), an emerging technique process data, successfully makes GNN possible handle Despite prosperous development vertical (VFGL), robustness VFGL against adversarial attack not been explored yet. Although numerous attacks centralized GNNs are proposed, their is challenged in...

Graph embedding learns low-dimensional representations for nodes or edges on the graph, which is widely applied in many real-world applications. Excessive graph mining promotes research of attack methods embedding. Most generate perturbations that maximize deviation prediction confidence. They are difficult to accurately misclassify instances into target label, and nonminimized more easily detected by defense methods. To address these problems, we propose Graphfool, a novel targeted label...

Graph neural networks (GNNs) have been successfully applied to a variety of graph-structure analysis tasks. Besides their outstanding performance, the explanation for GNNs’ predictions is also an inextricable problem, which hinders trust in GNNs under practical scenarios. Consequently, great efforts made interpreters understand behavior. However, existing works are still suffering two main problems: (i) explanation-shifting normal task - explanations provided by insufficient precisely...

Link prediction, inferring the undiscovered or potential links of graph, is widely applied in real-world. By facilitating labeled graph as training data, numerous deep learning based link prediction methods have been studied, which dominant accuracy compared with non-deep methods. However,the threats maliciously crafted will leave a specific backdoor model, thus when some examples are fed into it make wrong defined attack. It an important aspect that has overlooked current literature. In...

Graph neural network (GNN) with a powerful representation capability has been widely applied to various areas, such as biological gene prediction, social recommendation, etc. Recent works have exposed that GNN is vulnerable the backdoor attack, i.e., models trained maliciously crafted training samples are easily fooled by patched samples. Most of proposed studies launch attack using trigger either randomly generated subgraph (e.g., erd\H{o}s-r\'enyi backdoor) for less computational burden,...

Deep learning is effective in graph analysis. It widely applied many related areas, such as link prediction, node classification, community detection, and classification etc. Graph embedding, which learns low-dimensional representations for vertices or edges the graph, usually employs deep models to derive embedding vector. However, these are vulnerable. We envision that methods based on can be easily attacked using adversarial examples. Thus, this paper, we propose Graphfool, a novel...