Journal of Applied Science and Engineering

 Copyright The Author(s). This is an open access article distributed under the terms of the Creative Commons Attribution License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are cited.

Download Citation:  BibTeX | http://dx.doi.org/10.6180/jase.202611_34.018  

Download PDF

In globalized and digitally driven world, cross-border businesses face complex financial risks arising from differing regulations, geopolitical tensions, and unstable markets. Traditional risk management models struggle to process large-scale data in real time, limiting effective decision-making. This paper proposes an integrated big data framework for real-time risk analysis in cross-border operations. The framework incorporates Conditional Generative Adversarial Networks (cGANs) to address data imbalance, Long Short-Term Memory (LSTM) networks to forecast evolving risk patterns, and Extreme Value Theory (EVT) to quantify extreme losses through Value-at-Risk (VaR) and Conditional Value-at-Risk (CVaR). Reinforcement Learning (RL) is also included to optimize decision strategies such as hedging and capital allocation. With real-time monitoring and adaptive learning, the system supports proactive and scalable risk management. Experiments using the DataCo Smart Supply Chain Dataset show excellent performance, achieving 99.83% accuracy, 99.77% precision, 99.18% recall, 99.88% F1-score, and strong AUC results, validated further through VaR, CVaR, and Kupiec backtests.

Keywords: Conditional Generative Adversarial Networks, Extreme Value Theory, Graph Neural Networks, K-Nearest Neighbour and Long Short-Term Memory