Evaluating machine learning algorithms for intrusion detection systems using the dataset CIDDS-002
Quang-Vinh Dang · 2021
Intrusion detection systems play a vital role in protecting computer systems from external attacks. In recent years, many open-source intrusion datasets have been released to let researchers study and evaluate their detection methods. These datasets are presented with a comprehensive feature set. In this paper, we study the intrusion detection problem using a dataset CIDDS-002 that is provided with a small number of features. Our experimental using different classification algorithms including Logistic Regression, xgboost, Random Forest and catboost show that we can achieve very high predictive performance without using a lot of features generated in other datasets.