Can Large Language Models Improve the Adversarial Robustness of Graph Neural Networks?

TL;DR

This paper explores leveraging large language models to enhance the adversarial robustness of graph neural networks, proposing a novel framework that improves robustness against topology attacks despite some accuracy trade-offs.

Contribution

It introduces LLM4RGNN, a framework that distills GPT-4's inference capabilities into graph structure inference, significantly improving GNN robustness against topology perturbations.

Findings

LLM4RGNN consistently enhances GNN robustness across various models.

The framework maintains higher accuracy even with 40% perturbation ratio.

Using LLMs can effectively identify malicious edges and recover robust graph structures.

Abstract

Graph neural networks (GNNs) are vulnerable to adversarial attacks, especially for topology perturbations, and many methods that improve the robustness of GNNs have received considerable attention. Recently, we have witnessed the significant success of large language models (LLMs), leading many to explore the great potential of LLMs on GNNs. However, they mainly focus on improving the performance of GNNs by utilizing LLMs to enhance the node features. Therefore, we ask: Will the robustness of GNNs also be enhanced with the powerful understanding and inference capabilities of LLMs? By presenting the empirical results, we find that despite that LLMs can improve the robustness of GNNs, there is still an average decrease of 23.1% in accuracy, implying that the GNNs remain extremely vulnerable against topology attacks. Therefore, another question is how to extend the capabilities of LLMs on graph adversarial robustness. In this paper, we propose an LLM-based robust graph structure inference framework, LLM4RGNN, which distills the inference capabilities of GPT-4 into a local LLM for identifying malicious edges and an LM-based edge predictor for finding missing important edges, so as to recover a robust graph structure. Extensive experiments demonstrate that LLM4RGNN consistently improves the robustness across various GNNs. Even in some cases where the perturbation ratio increases to 40%, the accuracy of GNNs is still better than that on the clean graph. The source code can be found in https://github.com/zhongjian-zhang/LLM4RGNN.