DETECTINGWEB ATTACKS WITH END-TO-END DEEP LEARNING

Abstract

his study introduces an innovative deep learning-based method for identifying web-based attacks, addressing the limitations of conventional security tools. The proposed approach employs an end-to-end neural architecture capable of directly processing raw traffic data, learning intricate patterns of malicious activity without manual feature engineering. Through adaptive learning and real-time deployment, the framework can identify evolving cyber threats such as SQL Injection, Cross-Site Scripting, and Denial-of-Service attacks. The model’s high detection accuracy and minimal falsepositive rate demonstrate its potential as a robust, scalable security solution