Classification of Malicious Web Code Using Deep Learning
Manoj Kumar Selvam · NORMA · 2018
As the world moves towards web, there have been a lot of attacks which was carried out in web applications. Among those, we have the XSS (Cross-site scripting) attack which is considered as one of the top attack as per OWASP (Open Web-Application Security Project). The attacker uses a sophisticated method of injecting malicious code into the web application through web forms or request parameters, which is the stored in the server and later executed when the user visits the vulnerable page. In this paper we are using deep learning approach to identify the stored XSS vulnerabilities and to detect them, in order to prevent such malicious attacks. The paper discusses in detail the methodology used to apply deep machine learning methods and helps us identify malicious and non malicious web code. The experiment conducted using 11,000 labelled samples, provided the accuracy of 98% for the proposed architecture.