AI-Based Phishing Detection Techniques: A Comparative Analysis of Model Performance

Abstract

Abstract: Phishing attacks continue to pose significant threats to cybersecurity, targeting individuals, businesses, and organizations worldwide. In response, researchers and practitioners have turned to artificial intelligence (AI) techniques to enhance phishing detection capabilities. This paper presents a comparative analysis of AI-based phishing detection techniques, evaluating the performance of various machine learning (ML) and deep learning (DL) models in identifying phishing attempts.

The study explores a diverse range of features, including lexical, visual, and behavioral characteristics extracted from phishing emails and websites. Leveraging a dataset comprising real-world phishing instances, the performance metrics of different AI models are evaluated, including accuracy, precision, recall, and F1-score.

Furthermore, the paper investigates the robustness of AI-based phishing detection techniques against adversarial attacks and examines the generalization capabilities of models across different phishing scenarios and attack vectors.

The findings contribute to the understanding of the strengths and limitations of AI-based phishing detection approaches, offering insights into the most effective techniques for mitigating phishing threats in various contexts. Additionally, the study identifies areas for future research and development, such as the integration of ensemble learning methods and the incorporation of explainable AI techniques to enhance model interpretability and transparency.

Overall, this comparative analysis provides valuable guidance for cybersecurity practitioners and decision-makers in selecting and deploying AI-based phishing detection solutions to bolster their defenses against evolving cyber threats.