Intelligent Phishing Detection System Using Machine Learning | IJCT Volume 13 – Issue 2 | IJCT-V13I2P60

The rapidly evolving terror of phishing attacks presents a sobering challenge to protecting cybersecurity as they become increasingly creative in exploiting user trust. In the present study, an improved phishing detection framework was proposed that incorporates optimized feature selection combined with both machine and deep learning models to effectively distinguish between phishing and legitimate websites. The system being proposed has been developed in such a way that there is a step-by-step process for taking the data through the process of being prepared, breaking the data down into its individual parts, selecting the right parts of that data, creating and training a model, and finally, using that model to predict new data. Selecting the best features allows the system to reduce the amount of data involved (dimensionality), improve the speed of computation, and provide better generalization of the model. When evaluating different models (e.g., Graph Convolutional Networks, TabTransformer, Autoencoder, Feedforward Neural Networks, and Deep Neural Networks), it has been shown that deep learning methods using optimized feature sets are superior to traditional techniques. The experimental results confirm that the proposed framework provides improved detection accuracy, lower false positive rates and better robustness against previously unseen phishing attacks. Additionally, the deployment of the best model on a web interface demonstrates the practicability of the system in real-world settings. Overall, this research provides a scalable and effective method for detecting phishing and addresses major shortcomings of current approaches to enhance defenses against contemporary phishing attacks.