Related Experiment Video
Updated: Jul 13, 2026
08:20
Superior Auto-Identification of Trypanosome Parasites by Using a Hybrid Deep-Learning Model
Published on: October 27, 2023
Enhancing cybersecurity: A high-performance intrusion detection approach through boosting minority class recognition.
Chadia E L Asry1, Ibtissam Benchaji1, Samira Douzi1,2
1IPSS Laboratory, Faculty of Sciences, Mohammed V University, Rabat, Morocco.
Plos One
|March 28, 2025
Summary
This study introduces a novel approach for Intrusion Detection Systems (IDS) to improve the detection of rare cyberattacks. By using SHAP for feature selection and XGBoost, the system effectively identifies minority class threats like worms.
Area of Science:
- Cybersecurity
- Machine Learning
- Network Security
Background:
- Internet proliferation necessitates robust network security measures.
- Intrusion Detection Systems (IDS) struggle with identifying minority class attacks.
- Existing methods like class imbalance or resampling have limitations.
Purpose of the Study:
- To develop an effective method for detecting minority class attacks in network traffic.
- To enhance the precision and recall of Intrusion Detection Systems for rare threats.
- To address the limitations of current IDS methodologies in handling imbalanced datasets.
Main Methods:
- Utilized Shapley Additive Explanations (SHAP) for feature selection.
- Employed Recursive Feature Elimination with Cross-Validation (RFECV).
- Implemented XGBoost as the classification model.
Main Results:
- Achieved precision, recall, and F1-scores of 0.8095, 0.8293, and 0.8193 on the UNSW NB15 dataset.
- Demonstrated improved identification of minority class attacks, specifically 'worms'.
- Validated efficacy across CICIDS2019 and CICIoT2023 datasets.
Conclusions:
- The proposed SHAP-RFECV-XGBoost approach significantly enhances the detection of minority class cyberattacks.
- This method offers a more effective solution compared to traditional techniques for imbalanced network security data.
- The findings confirm the robustness and applicability of the approach across diverse datasets.
