Improve the performance of classification in computer networks itrusion detection with using features ranking

increasing range of computer networks and internet users followed by large amounts of the organization data, leading to an increased number of unauthorized activities. Since there is no complete secure system, intrusion detection and vulnerability analysis are very important. With respect to the increasing complexity and size of data, dimension reduction techniques with data mining have been proposed which indirectly have created a description that uses automated data analysis techniques. So in this paper, we aim to use Entropy Measures as feature selection and data mining techniques including K- nearest neighbor for intrusion detection. As results indicate, extended KNN algorithm with important features improved the Detection Rate and False Positive Rate in NSL-KDD dataset.

کلیدواژه ها:

intrusion detection system ، entropy measures ، data mining ، KNN