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

Adversarial Attacks on Clustering Coefficient in Complex Networks

Blindfolded Attackers Still Threatening: Strict Black-Box Adversarial Attacks on Graphs.

Vulnerability of Clustering under Node Failure in Complex Networks

Attacking the Core Structure of Complex Network

New attack strategies for complex networks

Adversarial Attacks to Scale-Free Networks: Testing the Robustness of Physical Criteria

The Dynamic Correlation Between Degree and Betweenness of Complex Network under Attack

Query-based Adversarial Attacks on Graph with Fake Nodes

Correlation of Centralities: A Study Through Distinct Graph Robustness

CORRELATION ANALYSIS OF ATTACK STRATEGIESAND ROBUSTNESS ANALYSIS ON THREE NETWORK MODELS

On Learning Cluster Coefficient of Private Networks

Attack Vulnerability of Network Controllability.

COREATTACK: Breaking Up the Core Structure of Graphs

Adversarial attack on BC classification for scale-free networks

The Robustness of Graph K-Shell Structure under Adversarial Attacks

Adversarial Learning for Robust Deep Clustering.

Measuring robustness of complex networks under MVC attack

SWCA: A Secure Weighted Clustering Algorithm in Wireless Ad Hoc Networks

Optimal Cost-based Strengthening of Complex Networks

Attack vulnerability of complex networks

Attacking The Assortativity Coefficient Under A Rewiring Strategy