EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC 4.0

A Comparative Study of Machine Learning Techniques for Intrusion Detection Systems

Munish Kumar

Journal of Contemporary Science and Technology Management · 2025

DOI: 10.66592/jcstm.01.04.02 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Journal of Contemporary Science and Technology Management, 2025. Verfasst von Munish Kumar. Der Volltext ist frei zugänglich (Open Access, CC BY-NC 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

The exponential growth in cyber attacks—now exceeding 56 million daily events—demands intelligent Intrusion Detection Systems (IDS) capable of identifying both known and novel threats with high accuracy and minimal false positives. This paper presents a comprehensive comparative study of machine learning techniques for network intrusion detection, evaluating eight algorithms spanning classical ML (Decision Trees, Random Forests, SVM, KNN), deep learning (LSTM, 1D-CNN, Transformer), and our proposed HybridNet architecture. HybridNet combines a 1D-CNN spatial feature encoder with a Bidirectional LSTM temporal sequence modeler through an attention-based fusion mechanism, enabling simultaneous capture of packet-level patterns and flow-level temporal dependencies. We conduct experiments on four benchmark datasets (NSL-KDD, UNSW-NB15, CIC-IDS2017, and a novel enterprise dataset with 12.8 million labeled flows) using rigorous cross-validation and time-series-aware splitting. HybridNet achieves 99.2% detection rate for known attacks and 78.4% for zero-day attacks with only 0.08% false positive rate, yielding an AUC of 0.997—surpassing all evaluated baselines. The proposed model processes 2,857 flows per second on a single GPU, meeting real-time requirements for enterprise network monitoring. We additionally present a feature importance analysis identifying the 15 most discriminative network features and a computational cost analysis demonstrating that HybridNet achieves optimal accuracy-efficiency trade-offs compared to all baselines.

Zitieren

APA

Kumar, M. (2025). A Comparative Study of Machine Learning Techniques for Intrusion Detection Systems. Journal of Contemporary Science and Technology Management. https://doi.org/10.66592/jcstm.01.04.02

Vancouver

Kumar M. A Comparative Study of Machine Learning Techniques for Intrusion Detection Systems. Journal of Contemporary Science and Technology Management. 2025. doi:10.66592/jcstm.01.04.02

BibTeX
@article{Kumar2025,
  author = {Kumar, Munish},
  title = {A Comparative Study of Machine Learning Techniques for Intrusion Detection Systems},
  journal = {Journal of Contemporary Science and Technology Management},
  year = {2025},
  doi = {10.66592/jcstm.01.04.02}
}
Kostenlos starten →