Abstract:While deep neural networks have achieved great success on the graph analysis, recent works have shown that they are also vulnerable to adversarial attacks where fraudulent users can fool the model with a limited number of queries. Compared with adversarial attacks on image classification, performing adversarial attack on graphs is challenging because of the discrete and non-differential nature of a graph. To address these issues, we proposed Cluster Attack, a novel adversarial attack by introducing a set of fake nodes to the original graph which can mislead the classification on certain victim nodes. Specifically, we query the victim model for each victim node to acquire their most adversarial feature, which is related to how the fake node’s feature will affect the victim nodes. We further cluster the victim nodes into several subgroups according to their most adversarial features such that we can reduce the searching space. Moreover, our attack is performed in a practical and unnoticeable manner: (1) We protect the predicted labels of nodes which we are not aimed for from being changed during attack. (2) We attack by introducing fake nodes into the original graph without changing existing links and features. (3) We attack with only partial information about the attacked graph, i.e., by leveraging the information of victim nodes along with their neighbors within k-hop instead of the whole graph. (4) We perform attack with a limited number of queries about the predicted scores of the model in a blackbox manner, i.e., without model architecture and parameters. Extensive experiments demonstrate the effectiveness of our method in terms of the success rate of attack. Department of Computer Science and Technology, Tsinghua University, China. Correspondence to: Zhengyi Wang <wangzy21@mails.tsinghua.edu.cn>. x x x

Using Bayesian inference for computing attack graph node beliefs

Dynamic Network Security Analysis Based on Bayesian Attack Graphs

A Probability-Based Approach to Attack Graphs Generation

Discovering Attack Scenarios Via Intrusion Alert Correlation Using Graph Convolutional Networks

Attack Graph Analysis for Network Anti-Forensics.

Evaluating Network Security With Two-Layer Attack Graphs

A New Method to Generate Attack Graphs

Bayesian Model for Tracing and Locating a Network Attack

SSIRI 2009 Short Paper A New Method to Generate Attack Graphs

A New Method of Network Risk Assessment Based on Bayesian Model

Research on Automatic Generation and Analysis Technology of Network Attack Graph

An Adjacency Matrixes-Based Model for Network Security Analysis

Network Security Situation Assessment Based on Attack Graph Techniques

Survey of Attack Graph Analysis Methods from the Perspective of Data and Knowledge Processing

Towards Reasonable Budget Allocation in Untargeted Graph Structure Attacks via Gradient Debias

A Full Bayesian Approach to Sparse Network Inference Using Heterogeneous Datasets

Bayesian inference of transition matrices from incomplete graph data with a topological prior

AI-based Attack Graph Generation

Inference Attacks Against Graph Neural Networks

Query-based Adversarial Attacks on Graph with Fake Nodes

Bayesian graph convolutional network with partial observations