Explainable AI for Transparent Phishing Email Detection
Y L D H Yakandawala, M K P Madushanka, W. M. K. S. Ilmini · 2024
Phishing attacks continue to pose significant threats by impersonating legitimate entities to extract sensitive information, often evading traditional security measures. This research addresses the need for advanced phishing detection systems capable of adapting to sophisticated tactics, focusing on leveraging machine learning models. The study evaluates various machine learning algorithms, including Support Vector Machines (SVM), Random Forest, and Logistic Regression, to identify the most effective model for phishing email detection. Explainable AI (XAI) techniques, such as Local Interpretable Model-Agnostic Explanations (LIME), are integrated to provide transparent, user-friendly explanations for model decisions. In testing, the SVM model achieved the highest accuracy (96%) and precision. User-friendly visualizations of the model's decision-making process significantly improved user trust. This paper highlights the potential of combining machine learning with XAI to enhance phishing detection, creating a robust, transparent system that empowers users to make informed decisions about email security.