EUVIMEDEUVIMED entdecken
Crossref

DAPTAD-FSDL: A Significant Feature-Driven Deep Learning Framework for Advanced Persistent Threat Detection in Networked Systems

Priya pathak, Priyamwada Sharma

Communications on Applied Nonlinear Analysis · 30.11.2025 · 32 (3) · 1197-1208

DOI: 10.52783/cana.v32.6567 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Communications on Applied Nonlinear Analysis, online erschienen am 30.11.2025. Verfasst von Priya pathak und 1 weiteren Autor:innen.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Advanced Persistent Threats (APTs) represent one of the most sophisticated and damaging forms of cyberattacks due to their stealthy, multi-stage, and long-duration nature. Traditional intrusion prevention systems and machine learning-based detection models often struggle to identify such threats accurately, primarily due to limited feature relevance, high false-positive rates, and poor generalization in dynamic network environments. To address these challenges, this paper proposes a Deep Learning-based Advanced Persistent Threat Attack Defense mechanism using Significant Features (DAPTAD-FSDL). The proposed framework integrates Z-score normalization for effective data preprocessing, Grey Wolf Optimizer (GWO)–based feature selection to identify the most informative attributes, and an attention-based CNN-BiLSTM architecture for robust APT classification. Furthermore, a Multi-Objective Golden Jackal Optimization Algorithm (MIGJOA) is employed to optimize the deep learning model hyperparameters, ensuring improved convergence, reduced overfitting, and enhanced generalization capability. The attention mechanism enables the model to focus on critical temporal and spatial patterns associated with multi-stage attack behaviors, significantly improving detection accuracy. The performance of the DAPTAD-FSDL framework is evaluated using the NSL-KDD dataset with a 70:30 training–testing split. Experimental results demonstrate that the proposed model achieves superior detection performance with an average accuracy of 96.67%, precision of 96.47%, recall of 96.67%, F-measure of 96.53%, and AUC score of 96.67%. Comparative analysis with existing machine learning and deep learning approaches further validates the effectiveness and robustness of the proposed framework. The results confirm that DAPTAD-FSDL provides a reliable and scalable solution for real-time APT detection in modern networked environments.

Zitieren

APA

pathak, P., & Sharma, P. (2025). DAPTAD-FSDL: A Significant Feature-Driven Deep Learning Framework for Advanced Persistent Threat Detection in Networked Systems. Communications on Applied Nonlinear Analysis, 32(3), 1197-1208. https://doi.org/10.52783/cana.v32.6567

Vancouver

pathak P, Sharma P. DAPTAD-FSDL: A Significant Feature-Driven Deep Learning Framework for Advanced Persistent Threat Detection in Networked Systems. Communications on Applied Nonlinear Analysis. 2025;32(3):1197-1208. doi:10.52783/cana.v32.6567

BibTeX
@article{pathak2025,
  author = {pathak, Priya and Sharma, Priyamwada},
  title = {DAPTAD-FSDL: A Significant Feature-Driven Deep Learning Framework for Advanced Persistent Threat Detection in Networked Systems},
  journal = {Communications on Applied Nonlinear Analysis},
  year = {2025},
  volume = {32},
  number = {3},
  pages = {1197--1208},
  doi = {10.52783/cana.v32.6567}
}
Kostenlos starten →