Enhancing Phishing Detection: A Machine Learning Approach with Feature Selection and Deep Learning Models

Phishing attacks have evolved as a major cybersecurity threat, exploiting user trust and compromising sensitive information. This study proposes an advanced phishing detection framework combining feature selection techniques with machine learning and deep learning models. Using a labeled dataset with the status field indicating legitimate or phishing websites, we evaluate and compare the performance of various models including Graph Convolutional Network (GCN), TabTransformer, Autoencoder, Feedforward Neural Network (FNN), and Deep Neural Network (DNN). By applying optimal feature selection, we enhance model performance, reduce computational complexity, and improve generalization. The system is implemented using Python and deployed with a Flask web interface styled with HTML and CSS, ensuring user-friendly interaction. Our results demonstrate that the integration of deep learning architectures with feature engineering significantly boosts phishing detection accuracy and robustness. This approach offers a scalable and effective solution to safeguard users against phishing threats in real-world applications.

Keywords: Phishing Detection, Feature Selection, GCN, TabTransformer, Autoencoder, FNN, DNN, Flask, Cybersecurity.