Enhanced Neural Network Based Combined Classification for Intrusion Detection System

  • Nusrath Unnisa A, Dr.Manjula Yerva, Dr. M. Z. Kurian


Communication is a method of conveying information between end users. In this digital era data communication plays a vital role where the information is transmitted through internet. All over the world, web applications are used to store sensitive and personal data. These issues make them an attractive target for malware that exploits vulnerabilities in order to acquire unauthorized access. In order to access a particular resource a unique address is used called URL (Uniform Resource locator). These URL’s can be attacked by an intruder to access the information stored on the client side or server by redirecting them. This paper proposes a model for intrusion detection system which is a combination of K-Nearest neighbour, Support vector machine and neural network in which the final output class is decided as Malicious or Benign based on the majority voting scheme. The work is carried out by using Cross Site scripting (XSS) attacks database and an alert is given to the end user that URL is malicious or benign.