A Comprehensive Investigation into Robust Malware Detection with Explainable Ai

Abstract

As cyber threats continue to evolve in complexity, the imperative for robust malware detection systems has never been more critical. This pa-per presents a comprehensive investigation into the integration of Explain-able AI (XAI) techniques to enhance the interpretability and transparency of malware detection models. The experimental study employs a meticu-lously designed methodology, incorporating a diverse dataset and cutting-edge XAI methods. Leveraging quantitative and qualitative analyses, our results demonstrate not only the efficacy of the malware detection system but also the invaluable insights provided by XAI. The findings shed light on the interpretability-accuracy trade-offs and highlight the potential of XAI in fostering trust and understanding within the cybersecurity land-scape. Through this exploration, we contribute to the ongoing discourse on advancing malware detection and underscore the transformative role of Explainable AI in fortifying digital defenses against contemporary cy-ber threats. The CICMalDroid dataset was utilized, and analysis was conducted using SHAP and LIME.

Keywords: malware, malware detection, explainable AI, SHAP, LIME, machine learning