GRAPH NEURAL NETWORKS FOR REAL-TIME MALWARE DETECTION IN ENTERPRISE ENVIRONMENTS

Li, XinYu and Roberts, Daniel and Bennett, Oliver (2025) GRAPH NEURAL NETWORKS FOR REAL-TIME MALWARE DETECTION IN ENTERPRISE ENVIRONMENTS. Social Science and Management, 2 (4). pp. 29-44. ISSN 3007-6854

Abstract

The escalating sophistication of malware threats poses unprecedented challenges to enterprise cybersecurity infrastructure. Traditional signature-based detection methods struggle to identify polymorphic and zero-day malware variants that continuously evolve to evade detection mechanisms. This research presents a comprehensive investigation into the application of Graph Neural Networks (GNNs) for real-time malware detection in enterprise environments. By leveraging the structural properties of malware represented as control flow graphs and function call graphs, GNN-based approaches can capture complex behavioral patterns that distinguish malicious software from benign applications. This study examines the theoretical foundations of graph-based malware representation, evaluates state-of-the-art GNN architectures including Graph Convolutional Networks and Graph Attention Networks, and proposes an integrated framework optimized for real-time detection in enterprise settings. Experimental evaluation demonstrates that the proposed approach achieves detection accuracy exceeding 96 percent while maintaining computational efficiency suitable for deployment in production environments. The findings indicate that GNN-based detection systems offer significant advantages over traditional machine learning methods, particularly in identifying previously unseen malware families through structural pattern recognition. This research contributes to the advancement of proactive cybersecurity measures by demonstrating the viability of graph-based deep learning for scalable, real-time threat detection in complex enterprise networks.