ML-Powered Intrusion Prevention for XSS Defense in Web Apps

This paper presents a system using machine learning to prevent XSS attacks on web applications. A deep neural network is built using the Keras library in Python and training using the CSIC 2010 dataset and data retrieved from pastebin.org. This neural network is then used to classify incoming HTTP requests according to whether they are XSS attacks or not. It was found that the system we developed blocks most attacks. A test using 25 XSS vectors showed an 84% success rate, which can be improved upon. The improvement, in our case, would come from developing better machine learning classifiers.