APDG-MO Adaptive Probabilistic Dependency Graphs With Memetic Optimization for Network Intrusion Detection
International Journal of Information Security and Privacy · 09.10.2026 · 20 (1) · 1-23
DOI: 10.4018/ijisp.424313 ↗Kurz gefasst
Fachartikel in der Zeitschrift International Journal of Information Security and Privacy, erschienen am 09.10.2026. Verfasst von Mahdi Ajdani. Die Arbeit nennt 37 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 3.0).
Abstract (Originalsprache)
Network intrusion detection systems (NIDSs) face significant challenges due to the high dimensionality of network traffic data, redundant features, and complex dependencies among traffic attributes. Traditional feature selection approaches often rely on static dependency assumptions and may fail to capture dynamic interactions between features during optimization. This paper proposes APDG-MO, an Adaptive Probabilistic Dependency Graph framework integrated with memetic optimization for effective feature selection in network intrusion detection. The proposed framework dynamically reconstructs probabilistic dependency relationships among features based on the current elite population and incorporates an attention-guided local refinement mechanism to improve the balance between exploration and exploitation. Unlike traditional tree-based dependency models, APDG enables more flexible representation of higher-order feature interactions. The proposed method is evaluated on three widely used benchmark datasets, including NSL-KDD, CICIDS2017, and UNSW-NB15, using XGBoost as the classification model. Experimental results demonstrate that APDG-MO achieves improved detection performance compared with baseline evolutionary and feature selection methods. On the CICIDS2017 dataset, the proposed framework achieves 92.8% accuracy and 91.6% Macro F1-score, while maintaining a compact feature representation. The results indicate that adaptive dependency modeling combined with attention-guided memetic optimization can provide an effective strategy for feature selection in modern network intrusion detection systems.
Zitieren
APA
Ajdani, M. (2026). APDG-MO Adaptive Probabilistic Dependency Graphs With Memetic Optimization for Network Intrusion Detection. International Journal of Information Security and Privacy, 20(1), 1-23. https://doi.org/10.4018/ijisp.424313
Vancouver
Ajdani M. APDG-MO Adaptive Probabilistic Dependency Graphs With Memetic Optimization for Network Intrusion Detection. International Journal of Information Security and Privacy. 2026;20(1):1-23. doi:10.4018/ijisp.424313
BibTeX
@article{Ajdani2026,
author = {Ajdani, Mahdi},
title = {APDG-MO Adaptive Probabilistic Dependency Graphs With Memetic Optimization for Network Intrusion Detection},
journal = {International Journal of Information Security and Privacy},
year = {2026},
volume = {20},
number = {1},
pages = {1--23},
doi = {10.4018/ijisp.424313}
}
EUVIMED