Large Language Models to Enhance Malware Detection in Edge Computing

The increasing sophistication of malware in recent years has underscored the need for more advanced detection approaches, particularly in edge computing environments. Traditional methods and machine learning models, have shown promise but face significant limitations in handling evolving malware types. To address these challenges, we explore the application of Large Language Models (LLMs) for malware detection in edge computing. LLMs offer enhanced pattern recognition, contextual understanding, and the ability to analyze complex data, making them well-suited for dynamic, resource-constrained edge environments. However, their use introduces challenges such as model size and computational demands. In this paper, we propose a vision architecture that leverages LLMs' strengths while addressing their limitations, providing a framework for continuous learning and adaptability. We discuss the architecture's potential impact on improving malware detection as well as future research directions in the field.