Analysis of a Network Intrusion Detection Method Based on MetaCost and Random Forest
This paper describes a network intrusion detection algorithm based on MetaCost-RF by combining the cost-sensitive learning method.The algorithm reduces the negative impact of unbalanced datasets on RF by introducing a cost matrix during RF training.Test validation of trained models on NSL-KDD.The results show that the accuracy of MetaCost-RF is improved by 5.16 percentage points compared with RF,and the recall rate of three minority classes is improved by 10.82,20.00 and 21.17 percentage points respectively.It is shown that the model is effective in enhancing the accuracy.Besides,it improving recall for a small number of classes of samples.
NETL
NSTL
万方数据
