Daily Arxiv

This is a page that curates AI-related papers published worldwide.
All content here is summarized using Google Gemini and operated on a non-profit basis.
Copyright for each paper belongs to the authors and their institutions; please make sure to credit the source when sharing.

An Explainable Transformer-based Model for Phishing Email Detection: A Large Language Model Approach

Created by
  • Haebom

Author

Mohammad Amaz Uddin, Md Mahiuddin, Iqbal H. Sarker

Outline

This paper presents the results of a study that optimizes and fine-tunes the Transformer-based DistilBERT model to improve phishing email detection performance in response to the increasing threat of phishing emails. We utilize preprocessing techniques to address the imbalanced dataset problem and experimentally demonstrate high accuracy. Furthermore, we ensure transparency by making the model's prediction process explainable through XAI techniques such as LIME and Transformer Interpret.

Takeaways, Limitations

•
Takeaways:
◦
We present the possibility of building an effective phishing email detection system using the Transformer-based DistilBERT model.
◦
Proposing an effective preprocessing technique for the imbalanced dataset problem.
◦
Improved reliability by making the model's prediction process explainable through XAI techniques.
◦
Proving the applicability of real systems by achieving high accuracy.
•
Limitations:
◦
Lack of detailed description of the characteristics of the dataset used.
◦
Lack of comparative analysis with other modern phishing detection techniques.
◦
Lack of discussion on the limitations of explanation and subjectivity of interpretation through XAI techniques.
◦
Lack of consideration of potential problems and solutions that may arise when applying the system to a real environment.
👍
Made with Slashpage