Enhanced Intrusion Detection System using Hybrid Machine Learning Approach
Pavan Singhal, Gajendra Singh · International journal of advanced research in computer science and electronics engineering · 2014
Modern business has run on technology and it is based on communication and consequently the gigantic speed of the today’s internet or communication is the cause of the advancement in telecommunication and semiconductor technologies together. Billions of users are accessing the internet hundreds of time in a day. Due to flexibility and ease of networking services security is the chief concern. To get protected Intrusion Detection and Preventions System are the best option to assure security. In this article Anomaly based IDPS has been proposed and evaluated using hybrid machine learning approach. Machine learning sub branch of the soft computing had evolved since last decade has present more promising solution in the field of the security (host and network). Various methods of machine learning have been tested to produces better results in detection of intrusive activities. Classification (KNN) and evidence theory (DS) is types of machine learning approach and support to provide better solution in this direction. Proposed method has adopted the idea of KNN and DS Theory to fasten the detection speed, achieving better efficiency and accuracy with low false positive and negative ratio. Obtained results have achieved the accuracy about 97.47% and false ratio has minimized and limited it to 1.2 and 1.3.