XSS Attack Detection using Convolution Neural Network

Abstract

Web applications are at a significant risk of being attacked by a diverse range of malicious actors because they are so widely used. These assaults can come from a variety of directions and can be of varying degrees of sophistication and severity, depending on the perpetrator. The development of Internet technology, together with advances in science and technology, has allowed it to permeate a number of different industries in today's society. However, with such rapid expansion comes the risk of compromised information security. In this group, the XSS vulnerability, which is often referred to as cross site scripting, has emerged as one of the most serious flaws in modern Internet applications. The most important task for network security is web attack detection. In order to address this challenging issue, this research investigates deep learning techniques and analyses them using convolutional neural networks. Convolutional neural networks are advantageous for XSS classification applications because of their architecture, which necessitates less pre-processing for feature extraction In this particular investigation, the Convolutional Neural Network (CNN) method was applied in order to categories and identify XSS scripts as either malicious or benign., and we almost exclusively used XSS script characters during feature creation. We achieved accuracy, precision, and recall values of 97.95, 99.30, and 96.66.