JAA IDS - FRAMEWORK DESIGN FOR AN EFFICIENT INTRUSION DETECTION SYSTEM

Abstract

Network security has become a very hot research area as its importance is heavily realized in various fields. Various mechanisms and tools are available to support this. But they do not meet the challenges imposed by fast growing technologies. Massive amounts of high dimensional data are one of the challenges. Data with a large number of features is entering and moving around the network. The Intrusion Detection System is a new mechanism that faces this challenge with the support of data mining and feature selection. In data mining, the ensemble is more preferable than a single method. In an ensemble, during the testing phase, all base classifiers are treated equally and individually participate and vote. To take a final decision, some extra effort has to be made. All these increase computation effort and time. To overcome these, this research paper proposes a new framework for intrusion detection systems using the Auto Bi-Level (ABL) classification technique with Double Filtering Fine Tuning–Ensemble Hybrid method.

Keywords

• network security
• features selection
• data mining
• intrusion detection system