Evolving AI-Based Malware Detection: A Hybrid Approach Combining Transfer Learning and Explainable AI

Abstract

In recent years, the proliferation of sophisticated malware has necessitated the development of advanced detection methods. This study introduces an innovative hybrid approach that combines Transfer Learning (TL) and Explainable AI (XAI) for malware detection. By leveraging TL, the model utilizes pre-trained neural networks, enabling it to efficiently recognize and adapt to new and evolving threats with minimal training data. XAI techniques are integrated to ensure transparency, allowing cybersecurity experts to understand and trust the model's decisions. This dual strategy not only enhances detection accuracy but also addresses the critical need for explainability in AI-driven security solutions. The proposed method is evaluated against various datasets, demonstrating its efficacy in detecting both known and emerging malware, while providing clear insights into its decision-making process. This hybrid approach represents a significant advancement in the field of AI-based cybersecurity, balancing performance with interpretability to better combat the dynamic nature of cyber threats.