Phishing Detection by integrating Machine Learning and Deep Learning
Soham Sawant, Rushabh Savakhande, Om Sankhe, Santosh Tamboli · 2024
Nowadays, where individuals are constantly utilizing the internet, phishing attacks are a threat to them. In such attacks, individuals’ personal information, financial information, etc. is gathered through phishing websites made by cybercriminals that mimic genuine websites. To address this, an effective safeguard system is required. Our approach is to develop a web browser extension using HTML, CSS, and JavaScript to distinguish phishing websites and a web application using Python Flask to obtain a comprehensive analysis of the URL. The extension acts as a shield to guard the users and keep them from getting to phishing websites. To boost performance, it employs deep learning methods like batch normalization and custom CNN (Convolutional Neural Network) and NLP (Natural Language Processing) techniques like tokenization and padding with traditional machine learning models like Random Forest. With a training accuracy of 97% and a validation accuracy of 96%, this combination not only effectively addresses the threat of phishing but also provides users with a complete web application that gives URL statistics, presenting them with a secure browsing experience.