Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{179820,
        author = {Golla Jashwanth Shiva and T.Raghavendra Gupta and Arakala Roshini and Gadila Srija Reddy and Gonam Pawan Kalyan},
        title = {HIERARCHICAL ADVERSARIAL ATTACKS AGAINST GRAPH NEURAL NETWORKS},
        journal = {International Journal of Innovative Research in Technology},
        year = {2025},
        volume = {11},
        number = {12},
        pages = {8635-8639},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=179820},
        abstract = {This paper studies the vulnerability of Graph 
Neural Networks against adversarial attacks by 
measuring the effect of both node feature and structure 
perturbations. This purpose is achieved by creating a 
synthetic graph dataset containing 2708 nodes, each 
with 1433 features, and 10556 edges. An instance of two
layer Graph Convolutional Network is trained on this 
synthetic dataset to solve a binary classification task. 
Two kinds of adversarial perturbations are introduced: 
(1) node feature perturbation-the features of selected 
nodes are subjected to adding some random noises; (2) 
graph structure perturbation-addition or deletion of 
edges. The model is then evaluated on the clean test set 
and then reevaluated after applying the adversarial 
perturbations. Results show a significant decrease in 
classification accuracy following the introduction of 
such attacks, thus emphasizing how vulnerable GNNs 
are to adversarial manipulation. This framework 
constitutes a valuable approach toward studying 
adversarial robustness within graph-based models and 
points toward more resilient architectures for practical 
applications with GNNs.},
        keywords = {Graph  Neural  Networks  (GNNs),  Adversarial Attacks, Node Feature Perturbation,  Graph  Structure  Perturbation,  Manipulation, Robustness Evaluation.},
        month = {May},
        }