EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Hybrid Machine Learning Approach for Enhancing Intrusion Detection Systems Using CICIDS2017 Dataset

Ara Zozan Miran, Govand Salih Kadir

Dasinya Journal for Engineering and Informatics · 08.05.2026 · 2 (2)

DOI: 10.65542/djei.v2i2.49 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Dasinya Journal for Engineering and Informatics, online erschienen am 08.05.2026. Verfasst von Ara Zozan Miran und 1 weiteren Autor:innen. Die Arbeit nennt 27 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Traditional Intrusion Detection Systems (IDSs) are often ineffective because they rely on signature-based methods, resulting in high false-positive rates. With the expansion of new Artificial Intelligence (AI), especially Machine Learning (ML) algorithms, used in IDS systems, it is possible to achieve high performance in detecting anomalous and known threats. For that reason, this study aims to use Machine Learning (ML) algorithms to learn different patterns of known and anomalous threats across the Data Link (Layer 2) and Network (Layer 3) characteristics of the OSI model, as a basis for adaptive systems for IDS frameworks. Two supervised models (Random Forest, XGBoost) and two unsupervised models (Isolation Forest, One-Class SVM) were compared on the CICIDS2017 online dataset. The Results indicate that XGBoost achieved 99.3% accuracy on known threats, while One-Class SVM achieved 92.1% accuracy for unknown threats. Later, model performance was evaluated using standard classification metrics with a paired t-test. The findings of this study show the importance of combining supervised and unsupervised ML algorithms as a hybrid system to detect, classify, and learn from known and anomalous threats on different layers of network traffic. As a result, the findings can serve as a base for the adaptive IDS systems that can learn features and achieve higher performance.

Zitieren

APA

Miran, A. Z., & Kadir, G. S. (2026). Hybrid Machine Learning Approach for Enhancing Intrusion Detection Systems Using CICIDS2017 Dataset. Dasinya Journal for Engineering and Informatics, 2(2). https://doi.org/10.65542/djei.v2i2.49

Vancouver

Miran AZ, Kadir GS. Hybrid Machine Learning Approach for Enhancing Intrusion Detection Systems Using CICIDS2017 Dataset. Dasinya Journal for Engineering and Informatics. 2026;2(2). doi:10.65542/djei.v2i2.49

BibTeX
@article{Miran2026,
  author = {Miran, Ara Zozan and Kadir, Govand Salih},
  title = {Hybrid Machine Learning Approach for Enhancing Intrusion Detection Systems Using CICIDS2017 Dataset},
  journal = {Dasinya Journal for Engineering and Informatics},
  year = {2026},
  volume = {2},
  number = {2},
  doi = {10.65542/djei.v2i2.49}
}
Kostenlos starten →