Phishing URL Detection using Deep Q-Networks with Convolutional Neural Networks

This research presents an innovative approach to phishing URL detection by integrating hybrid features with Deep Q-Networks and convolutional neural networks. This combined method leverages Deep Q-Networks to enhance decision-making policies and convolutional neural networks to extract critical features from URLs, enabling precise classification. This two-pronged model accurately identifies URLs, categorizing them effectively as either phishing or legitimate. Tested on a specialized dataset, the model shows outstanding performance, achieving an accuracy rate of 99.75%, thereby exceeding previous methods and demonstrating its suitability for deployment across various web environments. Our findings emphasize the scalability and effectiveness of our approach, signifying a major advancement in cybersecurity and phishing detection technologies.