EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

A Hybrid Deep Learning Framework for Intelligent Anomaly Detection and Automated Decision-Making in Large-Scale Computing Systems

Rajeev Kumar Dang, Nayani Sateesh, Dr. M. Kalaimani, Dr. Kanchan Doke, Dr. Yogesh Golhar, Nikita Katariya

Computer and Decision Making: An International Journal · 03.11.2026 · 3 (6S) · 2850-2859

DOI: 10.1922/comdem.v3i6s.19038 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Computer and Decision Making: An International Journal, online erschienen am 03.11.2026. Verfasst von Rajeev Kumar Dang und 5 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

The increasing complexity of large-scale computing environments has created a need for intelligent anomaly-detection mechanisms capable of identifying rare and context-dependent deviations in dynamic temporal data. This study presents an intelligent anomaly-detection framework based on temporal feature engineering and unsupervised machine learning to establish a baseline for future hybrid deep-learning and automated decision-making systems. The analysis comprised 5,159 complete hourly observations derived after temporal aggregation and lag construction. Predictor variables included observed value, hour, day, calendar day, month, seven-period rolling mean, and lagged value. An Isolation Forest model with 200 estimators, a contamination rate of 0.005, and a maximum sample fraction of 0.70 was employed for anomaly identification. The analyzed series showed substantial temporal variability, with a mean value of 15,138.66 and a range from 9.50 to 37,204.50. The model detected 26 anomalous observations, corresponding to 0.504% of the analytical sample. Most anomalies were temporally concentrated, with 24 occurring within a single month, while the strongest anomaly score was -0.0779. The findings indicate that contextual and sequential features are important for distinguishing genuine abnormalities from routine temporal fluctuations. The framework provides a reproducible baseline for extending anomaly detection toward hybrid LSTM, CNN-LSTM, autoencoder, transformer, and automated decision-support architectures in complex computing environments.

Zitieren

APA

Dang, R. K., Sateesh, N., Kalaimani, D. M., Doke, D. K., Golhar, D. Y., & Katariya, N. (2026). A Hybrid Deep Learning Framework for Intelligent Anomaly Detection and Automated Decision-Making in Large-Scale Computing Systems. Computer and Decision Making: An International Journal, 3(6S), 2850-2859. https://doi.org/10.1922/comdem.v3i6s.19038

Vancouver

Dang RK, Sateesh N, Kalaimani DM, Doke DK, Golhar DY, Katariya N. A Hybrid Deep Learning Framework for Intelligent Anomaly Detection and Automated Decision-Making in Large-Scale Computing Systems. Computer and Decision Making: An International Journal. 2026;3(6S):2850-2859. doi:10.1922/comdem.v3i6s.19038

BibTeX
@article{Dang2026,
  author = {Dang, Rajeev Kumar and Sateesh, Nayani and Kalaimani, Dr. M. and Doke, Dr. Kanchan and Golhar, Dr. Yogesh and Katariya, Nikita},
  title = {A Hybrid Deep Learning Framework for Intelligent Anomaly Detection and Automated Decision-Making in Large-Scale Computing Systems},
  journal = {Computer and Decision Making: An International Journal},
  year = {2026},
  volume = {3},
  number = {6S},
  pages = {2850--2859},
  doi = {10.1922/comdem.v3i6s.19038}
}
Kostenlos starten →