A Scalable Feature Optimization Framework for Intrusion Detection Systems Using GA-SUS and SMOTE
Communications on Applied Nonlinear Analysis · 25.02.2025 · 32 (1s) · 970-978
DOI: 10.52783/cana.v32.7014 ↗Kurz gefasst
Fachartikel in der Zeitschrift Communications on Applied Nonlinear Analysis, online erschienen am 25.02.2025. Verfasst von Radharani Akula und 1 weiteren Autor:innen.
Abstract (Originalsprache)
Intrusion Detection Systems (IDS) play a vital role in protecting modern network infrastructures against evolving cyber threats. However, the high dimensionality and redundancy of network traffic features often reduce detection efficiency and increase computational complexity. This paper proposes an optimized feature selection framework for IDS using a Genetic Algorithm (GA) integrated with Stochastic Universal Sampling (SUS) to identify the most relevant network traffic features from the CICIDS2017 dataset. Initially, preprocessing techniques including data cleaning, normalization, and label encoding were applied to improve data quality. To address severe class imbalance in the dataset, the Synthetic Minority Oversampling Technique (SMOTE) was employed. The proposed GA-SUS model performs intelligent feature subset optimization by selecting informative features while minimizing redundancy. A Random Forest classifier was then trained using the selected optimal feature subset for intrusion detection. Experimental results demonstrate that the proposed framework successfully reduced the feature space from 77 to 32 features while achieving a high classification accuracy of 97.76% on multiclass intrusion detection tasks. The obtained results indicate that the integration of Genetic Algorithm and Stochastic Universal Sampling significantly improves IDS efficiency, reduces computational overhead, and maintains robust detection performance. The proposed method can effectively support real-time cybersecurity applications requiring accurate and scalable intrusion detection mechanisms.
Zitieren
APA
Akula, R., & Kumar, G. N. (2025). A Scalable Feature Optimization Framework for Intrusion Detection Systems Using GA-SUS and SMOTE. Communications on Applied Nonlinear Analysis, 32(1s), 970-978. https://doi.org/10.52783/cana.v32.7014
Vancouver
Akula R, Kumar GN. A Scalable Feature Optimization Framework for Intrusion Detection Systems Using GA-SUS and SMOTE. Communications on Applied Nonlinear Analysis. 2025;32(1s):970-978. doi:10.52783/cana.v32.7014
BibTeX
@article{Akula2025,
author = {Akula, Radharani and Kumar, GS Naveen},
title = {A Scalable Feature Optimization Framework for Intrusion Detection Systems Using GA-SUS and SMOTE},
journal = {Communications on Applied Nonlinear Analysis},
year = {2025},
volume = {32},
number = {1s},
pages = {970--978},
doi = {10.52783/cana.v32.7014}
}
EUVIMED