Research map / Adversarial robustness and security / Attacks on text and graphs

GOttack: Universal Adversarial Attacks on Graph Neural Networks via Graph Orbits Learning

ICLR 2025 · Adversarial robustness and security › Attacks on text and graphs · approach: Graph neural network robustness

Read the paperCheck your idea against 108,967 papers

AI-extracted summary. Written by AI from the paper's abstract only. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Proposes GOttack, a universal adversarial attack on GNN node classifiers that exploits graph orbit (topological) structure to perturb graphs and degrade predictions.

Problem

Graph neural networks for node classification are vulnerable to adversarial structure perturbations that mislead predictions.

Approach

An adversarial attack framework that leverages graph orbits (topological structure patterns) to learn universal perturbations that undermine GNN predictions.

Closest papers by meaning

• Adversarial Robust Generalization of Graph Neural Networks (ICML 2025 · similarity 0.90)
• A Simple and Yet Fairly Effective Defense for Graph Neural Networks (AAAI 2024 · similarity 0.89)
• Bounding the Expected Robustness of Graph Neural Networks Subject to Node Feature Attacks (ICLR 2024 · similarity 0.89)
• Value at Adversarial Risk: A Graph Defense Strategy against Cost-Aware Attacks (AAAI 2024 · similarity 0.89)
• TRACI: A Data-centric Approach for Multi-Domain Generalization on Graphs (AAAI 2025 · similarity 0.89)
• When Witnesses Defend: A Witness Graph Topological Layer for Adversarial Graph Learning (AAAI 2025 · similarity 0.89)