ABSTRACT

Due to continuous growth of the internet technology, there is need to establish security mechanism. So for achieving this objective various NIDS has been propsed. Datamining is one of the most effective techniques used for intrusion detection. This work evaluates the performance of unsupervised learning techniques over benchmark intrusion detection datasets. The model generation is computation intensive, hence to reduce the time required for model generation various feature selection algorithm has been used. Problems with k-mean clustering are hard cluster to class assignment, class dominance, and null class problems. From experimental results it is observed that for 2 class datasets filtered fuzzy random forest dataset gives the better results. It is having 99.2% precision and 100% recall, So it can be summarize that proposed statistical model is giving better performance better results than existing clustering algorithm.