EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

A Real-Time AI-Based Cyber Threat Intelligence Framework for Securing Modern Banking Infrastructure

Kunal Chatterjee

Journal of Intelligent Decision Making and Information Science · 02.10.2026 · 3 (15s) · 2278-2292

DOI: 10.17762/jidmis.v3.4810 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Journal of Intelligent Decision Making and Information Science, online erschienen am 02.10.2026. Verfasst von Kunal Chatterjee. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Banks are undergoing rapid banking infrastructure transformation, growing the cyber-attack surface as well as increasing their need for intelligent, real-time capabilities to address emerging threats against traditional signature and rule-based detection. Previous study tends to focus on individual threat detection, fraud detection, or cloud security, but hasn't integrated behavioural analysis, temporal learning, anomaly detection, threat classification and actionable cyber threat intelligence into a common banking framework. This study proposes a real-time “AI-based Cyber Threat Intelligence (AI-CTI)” framework that includes data preprocessing, feature selection, temporal sequence construction, temporal learning by LSTM, autoencoder-based anomaly detection, threat classification and Cyber Threat Intelligence generation. A dataset of banking transactions underwent missing-value treatment, the removal of outliers via the Z score method, one-hot encoding, Min-Max normalization and fifteen features were selected that were relevant to the dataset. These processes were followed by temporal sequence modelling using a sliding window of the previous 10 observations. Imputing missing values using the median got rid of the figures that were missing in the datasets, ranging from about 3 to 8%, while the Z-score analysis identified about 2.5% of the 1,000 simulated values as outliers. The validation error for the LSTM was around 94% after 30 epochs and the validation error for the autoencoder was its RMSE of 0.09 for distinguishing anomalous behaviour. The entire framework was successful in achieving 96.4% accuracy, 94.8% precision, 95.6% recall and 95.2% F1-score. The insights gained from this study highlight the ability to combine temporal and behavioural AI approaches to support real-time threat detection and generation of actionable intelligence in banking.

Zitieren

APA

Chatterjee, K. (2026). A Real-Time AI-Based Cyber Threat Intelligence Framework for Securing Modern Banking Infrastructure. Journal of Intelligent Decision Making and Information Science, 3(15s), 2278-2292. https://doi.org/10.17762/jidmis.v3.4810

Vancouver

Chatterjee K. A Real-Time AI-Based Cyber Threat Intelligence Framework for Securing Modern Banking Infrastructure. Journal of Intelligent Decision Making and Information Science. 2026;3(15s):2278-2292. doi:10.17762/jidmis.v3.4810

BibTeX
@article{Chatterjee2026,
  author = {Chatterjee, Kunal},
  title = {A Real-Time AI-Based Cyber Threat Intelligence Framework for Securing Modern Banking Infrastructure},
  journal = {Journal of Intelligent Decision Making and Information Science},
  year = {2026},
  volume = {3},
  number = {15s},
  pages = {2278--2292},
  doi = {10.17762/jidmis.v3.4810}
}
Kostenlos starten →