An effective model for anomaly IDS to improve the efficiency

We all know that the information passed through internet is in terms of packets. The alerts produced by all the existing intrusion detection systems are false alerts which can cause to decrease the efficiency and the accuracy is also low. The alerts generated by all the existing intrusion detection systems are isolated alerts and they will focuses on low-level attacks. So in this research paper diverse data mining techniques are used to reduce false alarm rate in intrusion detection system and for improving its' efficiency. The techniques which are used here are K-Nearest Neighbor, K-Means and Decision Table Majority rule based. This research operates on the KDD'99 dataset for diverse invasion recognition systems. In this paper we first apply the grouping on the KDD'99 dataset then it can be classified into four categories as U2R, R2L, DoS and Probe. The important goal of this paper is to decrease the false positive rate of IDS and attempt to improve its efficiency.