Intrusion Detection System Using Data Mining Techniques
Nowadays the information maintained inside the servers is the most valuable asset to the Organizations. In order to protect the information in servers the Intrusion detection System (IDS) has come into picture. In this paper, we are designing a methodology to identify intruders effectively, also focused to reduce the searching time for the intruder. In our research, we used the k-means algorithm and Jaccard distance similarity measure. The 1998 DARPA dataset is used to classify the users as intruders or normal users. Each document of the dataset is converted into 32-bit binary signature using Hashing and Superimposed Coding techniques. The binary database is clustered and evaluated the intrusions. The IDS is detecting the unknown intruders with accuracy of 76.4% whereas it is detecting known users with 99.9% accuracy. The clustering technique improves the performance of our Intrusion Detection System.