ABSTRACT
Information safety, security and stability of the systems have been a serious is from and individual to an enterprise and are also remained a serious concern
globally. With the rapid growth of Internet the complexity, availabili y, and accessibility have helped to raise the safekeeping risk of information systems enormously. Accessing any network there always has been a threat of attac that
compromise confidentiality, proper working of the system and resources. o
safeguard against all possible intrusions there has been number of different ways like firewall, antivirus and Intrusion Detection Systems (IDS). To make the information
system safe and secure, the intrusion detection acts as critical component. KDD intrusion detection dataset offers labeled data for the scientists and researchers, choosing most important features or patterns from input dataset makes problem simpler, faster and acquires much more accuracy towards threat detection. In our work we demonstrate the problem of recognizing most important input patterns to design a more efficient Intrusion Detection System (IDS). Consequently, removal of irrelevant or unimportant inputs makes the problem of detecting a threat simpler, faster, and accurate. It has been an important issue in the domain of intrusion detection that features selection and ranking must be made accordingl y; because, it is the only way left to detect intrusion accurately and efficiently. We implement the
procedure to remove one feature at a time to run experiments on upport Vector Machine (SVM) to grade the significance of the features for the KDD collected intrusion dataset. It is revealed that SVM based IDSs utilizin g a lesser number of features could give improved and efficient performance ce.

xiii