Intrusion Detection Model Based on Rough Set and Random Forest

Currently, redundant data affects the speed of intrusion detection. Many intrusion detection systems (IDS) have low detection rates and high false alert rate. Focusing on these weakness, a new intrusion detection model based on rough set and random forest (RSRFID) is designed. In the intrusion detection model, rough set (RS) is used to reduce the dimension of redundant attributes; the algorithm of decision tree (DT) is improved; a random forest (RF) algorithm based on attribute significances is proposed. Finally, the simulation experiment is given on NSL-KDD and UNSW-NB15 dataset. The results show that attributes of different types of datasets are reduced using RS, the detection rate of NSL-KDD is 93.73%, the false alert rate is 1.02%, the detection rate of NSL-KDD is 98.92%, and the false alert rate is 2.92%.