Poster: Enhancing GNN Robustness for Network Intrusion Detection via Agent-based Analysis

2025-06-25

AI Summary

Graph Neural Networks (GNNs) for IoT intrusion detection suffer from insufficient robustness due to distribution shift and real-world adversarial attacks. Method: This paper proposes the first LLM-augmented GNN framework, wherein a large language model acts as a security expert agent to systematically model and analyze both black-box and white-box adversarial attack scenarios. The framework integrates network-flow graph modeling, a physically collected real-world adversarial dataset from a testbed, and an end-to-end training mechanism to enhance generalization. Contribution/Results: Experiments demonstrate that the proposed method significantly improves GNN-based detector stability and accuracy under diverse realistic adversarial attacks, achieving an average 12.7% F1-score gain. It constitutes the first empirical validation of LLMs as an interpretable, collaborative, defense-aware layer in network intrusion detection systems (NIDS).

Technology Category

• Machine Learning: Adversarial Learning & Robustness
• Computer Vision: Adversarial Attacks & Robustness
• Multiagent Systems: Adversarial Agents

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Search Tool Learning with LLM: Teaching LLMs to invoke search and make use of retrieved information

Abstract

Graph Neural Networks (GNNs) show great promise for Network Intrusion Detection Systems (NIDS), particularly in IoT environments, but suffer performance degradation due to distribution drift and lack robustness against realistic adversarial attacks. Current robustness evaluations often rely on unrealistic synthetic perturbations and lack demonstrations on systematic analysis of different kinds of adversarial attack, which encompass both black-box and white-box scenarios. This work proposes a novel approach to enhance GNN robustness and generalization by employing Large Language Models (LLMs) in an agentic pipeline as simulated cybersecurity expert agents. These agents scrutinize graph structures derived from network flow data, identifying and potentially mitigating suspicious or adversarially perturbed elements before GNN processing. Our experiments, using a framework designed for realistic evaluation and testing with a variety of adversarial attacks including a dataset collected from physical testbed experiments, demonstrate that integrating LLM analysis can significantly improve the resilience of GNN-based NIDS against challenges, showcasing the potential of LLM agent as a complementary layer in intrusion detection architectures.

Problem

• Enhancing GNN robustness against adversarial attacks in NIDS
• Addressing distribution drift and unrealistic synthetic perturbations in evaluations
• Integrating LLM agents to improve GNN-based intrusion detection resilience

Innovation

• LLMs simulate cybersecurity expert agents
• Agents analyze and mitigate adversarial graph elements
• Framework tests GNN robustness with realistic attacks