Precision enhancement of Intrusion detection system through outlier detection and feature classification
Abstract: The day by day beleaguered network attacks is gradually growing and evolving, forcing businesses to overhaul their network security systems due to feasible data and capital losses. Intrusion Detection Systems is a significant factor for almost any security system. The key feature of IDS is the dynamic detection of illegal access that tries to negotiation the privacy, ease of use and Integrity of computer or computer networks. A lot of researchers have previously developed protection and advanced technique to discover technologies to sense cyber attacks with all DARPA 1998 dataset for Intrusion Detection and improved versions of this KDD Cup'99, NSL-KDD Cup and GureKDDcup data set.In this research, we estimate the use of five ML categorization algorithm to deal with the attack classification difficulty. They are SVM, Naive Bayes, KNN and the Decision Tree based C4.5 (J48) and Random Forest Algorithm.