Network intrusion detection using data dimensions reduction techniques

Abstract Due to the increasing growth of Internet and its widespread application, number attacks on network has also increased. Therefore, maintaining security using intrusion detection systems is critical importance. The connection between devices leads a large data being generated saved. era “big data” emerges over time. This paper presents new method for selecting effective features based concept fuzzy numbers scoring methods correlation feature selection systems. goal this present approach reducing size In method, eliminate inefficient reduce dimensions, are defined as number, heuristic function correlation-based algorithm expressed triangular membership function. To evaluate proposed it then compared previous methods. results show that selects several less than conventional with higher rate. two datasets. evaluated validated KDD Cup, NSL-KDD CICIDS achieved accuracy 99.9% which 96.01% CFS method.