Phishing Detection Using Deep Learning for Web-Browser

Abstract

This paper suggests an end-to-end phishing detection system that includes web and email security hybrid deep learning models. For URL-based threats, URL format and page title are checked by a Char-CNN model, and a multilayer-perceptron deals with structured web attributes. In tandem, the LSTM-based approach is employed for email phishing detection, temporal and contextual extraction patterns through deep sequential learning. Both systems are deployed with scalable RESTful APIs for real-time threat classification. The shared platform emphasizes low latency, model scalability, and real-world integration into current infrastructures.