Research Challenges and Performance of Clustering Techniques to Analyze NSL-KDD Dataset
Yadav Prasad Raiwani, Shailesh Singh Panwar · 2014
Due to different malicious activities over Internet, there are major challenges to the research community as well as to the corporations. Many data mining techniques have been adopted for this purpose i.e. classification, clustering, association rule mining, regression, visualization etc. For this purpose clustering provides a better representation of network traffic in order to identify the type of data flowing through network. Clustering algorithms have been used most widely as an unsupervised classifier to organize and categorize data. In this paper we have analyzed four different clustering algorithms using NSL-KDD dataset. We tried to cluster the dataset in two classes i.e. normal and anomaly, using Kmeans, EM, DB clustering and COBWEB. The main objective of this evaluation is to determine the class labels of different type of data present in intrusion detection dataset and to find out efficient clustering algorithm. The results of the evaluation are compared and challenges faced in these evaluations are than discussed.