Regresar

STUDY OF DATA MINING TECHNIQUES FOR THE DETECTION OF ATTACKS IN THE NSL-KDD DATA SET

Abstract:

The present investigation begins with a referential study of similar works based on scenarios for the detection of intruders in the network applying data mining techniques in a data set with attributes referring to network connections. In particular, the NSL-KDD data set has been taken. Then it is intended to replicate results of previous investigations where applying classifying algorithms determines if a connection is normal or a network attack. After this, it is complemented with the application of new classification algorithms to obtain better results, as well as new attribute selector algorithms in order to reduce or change certain attributes to obtain similar results. Finally, a selection of attributes based on the frequency of appearance in previous subsets is proposed. For the comparison of the results, the percentages of successes and construction time of the models of each applied algorithm have been taken.