Lightweight Cloud-Based Phishing Email Detection using BERT and Deep Learning

The proliferation of phishing attacks in the digital era poses a significant threat to cybersecurity. This paper introduces a novel approach for phishing email detection by leveraging the Bidirectional Encoder Representations from Transformers (BERT) model and deep learning layers, implemented in a cloud-based environment. Our model used BERT to extract the features from the input mails and then a dense layer deep learning model is used to classify mail as ’Safe’ or ’Phishing’. As we used BERT only for feature selection and linear deep learning model for classification, our model is lightweight as compared to other phishing detection models. The model was trained and validated on a comprehensive dataset, demonstrating an overall accuracy of 98%, with high precision and recall metrics, further corroborated by F1-scores and a confusion matrix analysis. The integration of our proposed model with cloud computing not only improved the model’s efficiency and scalability but also enhanced its applicability in real-time phishing detection. The results of our study indicate a substantial advancement in phishing email detection, showcasing the potential of combining cutting-edge AI techniques with cloud technology in cybersecurity.