Addressing the class imbalance problem in network intrusion detection systems using data resampling and deep learning

Abstract Network intrusion detection systems (NIDS) are the most common tool used to detect malicious attacks on a network. They help prevent ever-increasing different and provide better security for NIDS classified into signature-based anomaly-based detection. The type of is which based machine learning models able with high accuracy. However, in recent years, has achieved even results detecting already known novel adoption deep models. Benchmark datasets try simulate real-network traffic by including more normal samples than attack samples. This causes training data be imbalanced difficulties certain types NIDS. In this paper, resampling technique proposed Adaptive Synthetic (ADASYN) Tomek Links algorithms combination mitigate class imbalance problem. model evaluated benchmark NSL-KDD dataset using accuracy, precision, recall F-score metrics. experimental show that binary classification, method improves performance outperforms state-of-the-art an accuracy 99.8%. multi-class were also improved, outperforming 99.98%.