Artificial Intelligence-Based Anomaly Detection for Environmental and Financial Risk Management in Banking

Aftab Hossain Sazu; Mst Anupa Nasrin  Khan; Shakila  Sultana; Masrufa Tasnim; Tanaya  Jakir; Abidul Alam; Sk Md Zubair; Md  Himeluzzaman

doi:10.29284/bc1a6813

Authors

Aftab Hossain Sazu Master of Business Administration (MBA), College of Business Administration, Central Michigan University, USA.
Mst Anupa Nasrin Khan Master of Science in Zoology (Entomology Focus), The University of Rajshahi, Bangladesh.
Shakila Sultana Masters in Accounting (M.A), National University, Bangladesh.
Masrufa Tasnim Master of Business Administration, University of Dhaka, Dhaka, Bangladesh.
Tanaya Jakir MS in Business Analytics, Trine University, USA.
Abidul Alam Master of Science in Information Technology, Washington University of Science and Technology, USA.
Sk Md Zubair Master of Science in Engineering Management, Trine University, Arizona, USA.
Md Himeluzzaman M.Sc. in Information Technology Systems and Management, Washington University of Science and Technology, VA, USA.

DOI:

https://doi.org/10.29284/bc1a6813

Keywords:

anomaly detection; outlier identification; ESG; environmental risk; autoencoder; explainable AI; banking risk governance; financial institutions.

Abstract

Financial institutions are more and more exposed to ecological risk through the environmental behavior of firms they finance, hold on their balance sheet, or insure. This makes the identification of those few counterparties that deviate on environmental indicators as much a governance problem as it is a modelling one. We provide a framework for the identification of outliers in firm environmental and financial profiles relative to their peers through artificial intelligence so that these firms can be flagged for review by a banking risk governance process. Using a panel of 1,000 firms that combines financial variables with environmental, social, and governance (ESG) indicators and physical externalities (carbon emissions, water usage, energy consumption) observed for the period 2015–2025 (11.000 firm year records), we test four classical detectors Isolation Forest, Local Outlier Factor, One-Class Support Vector Machine and Elliptic Envelope) against two deep learning detectors using a GPU: one is a deep autoencoder and the other is a variational autoencoder. We evaluate precision, recall, F1, ROC AUC, and PR AUC, along with detector agreement, bootstrap stability, and contamination rate sensitivity, as there are no ground-truth anomaly labels. The highest-ranking quality (ROC AUC 0.990) and the best F1 score (0.791) at a budget of five percent contamination per tested point are achieved by the deep autoencoder and Local Outlier Factor, respectively. Together, the ensemble of all detectors provides stable rankings across bootstrap resamples (mean Spearman 0.949), focusing flags on sectors of high gas emission. Reconstruction error attribution explains each feature-level flag, allowing the analyst to have an auditable reason for escalation. The framework serves as an early warning mechanism for the governance of ecological threats at financial institutions.

References

[1] Breunig, M. M., Kriegel, H.-P., Ng, R. T., & Sander, J. (2000). LOF: Identifying density-based local outliers. Proceedings of the ACM SIGMOD International Conference on Management of Data, 93 to 104.

[2] Liu, F. T., Ting, K. M., & Zhou, Z.-H. (2008). Isolation Forest. Proceedings of the IEEE International Conference on Data Mining (ICDM), 413 to 422.

[3] Scholkopf, B., Platt, J. C., Shawe-Taylor, J., Smola, A. J., & Williamson, R. C. (2001). Estimating the support of a high-dimensional distribution. Neural Computation, 13(7), 1443 to 1471.

[4] Rousseeuw, P. J., & Van Driessen, K. (1999). A fast algorithm for the minimum covariance determinant estimator. Technometrics, 41(3), 212 to 223.

[5] Hinton, G. E., & Salakhutdinov, R. R. (2006). Reducing the dimensionality of data with neural networks. Science, 313(5786), 504 to 507.

[6] Kingma, D. P., & Welling, M. (2014). Auto-encoding variational Bayes. Proceedings of the International Conference on Learning Representations (ICLR).

[7] An, J., & Cho, S. (2015). Variational autoencoder based anomaly detection using reconstruction probability. Special Lecture on IE, Seoul National University.

[8] Ruff, L., Vandermeulen, R., Goernitz, N., Deecke, L., Siddiqui, S. A., Binder, A., Muller, E., & Kloft, M. (2018). Deep one-class classification. Proceedings of the International Conference on Machine Learning (ICML), 4393 to 4402.

[9] Chandola, V., Banerjee, A., & Kumar, V. (2009). Anomaly detection: A survey. ACM Computing Surveys, 41(3), 1 to 58.

[10] Lundberg, S. M., & Lee, S.-I. (2017). A unified approach to interpreting model predictions. Advances in Neural Information Processing Systems (NeurIPS), 30.

[11] Pedregosa, F., Varoquaux, G., Gramfort, A., Michel, V., Thirion, B., Grisel, O., et al. (2011). Scikit-learn: Machine learning in Python. Journal of Machine Learning Research, 12, 2825 to 2830.

[12] Paszke, A., Gross, S., Massa, F., Lerer, A., Bradbury, J., Chanan, G., et al. (2019). PyTorch: An imperative style, high-performance deep learning library. Advances in Neural Information Processing Systems (NeurIPS), 32.

[13] ESG and Financial Performance Dataset. Kaggle open dataset (simulated ESG and financial performance of global companies, 2015 to 2025).

[14] Alim, M. A., Rahman, M. R., Arif, M. H., & Hossen, M. S. (2020). Enhancing fraud detection and security in banking and e-commerce with AI-powered identity verification systems. World Journal of Advanced Research and Reviews, 2035.

[15] Rahman, M. R., Rahman, M., Rasul, I., Arif, M. H., Alim, M. A., Hossen, M. S., & Bhuiyan, T. (2024). Lightweight Machine Learning Models for Real-Time Ransomware Detection on Resource-Constrained Devices. Journal of Information Communication Technologies and Robotic Applications, 15(1), 17-23.

[16] Rahman, M., Arif, M. H., Alim, M. A., Rahman, M. R., & Hossen, M. S. (2021). Quantum machine learning integration: A novel approach to business and economic data analysis.

[17] Tohfa, N. A., Hossain, I., Zareen, S., Rasul, I., Hossen, M. S., & Rahman, M. (2021). Adversarial Cognition Machine Learning at the Frontlines of Cyber Warfare. World Journal of Advanced Research and Reviews, 12(02), 722-729.

[18] Hossain, I., Islam, M. S., Sultana, R., & Rahman, M. R. (2022). IoT based home automation system using renewable energy. American Journal of Agricultural Science, Engineering, and Technology.

[19] Rahaman, M. M., Biswas, B., & Hossain, M. S. (2022). Dynamic task prioritization in Meta-GNN (graph neural networks) for fraud detection: A meta-reinforcement learning approach with adaptive graph sparsification. International Journal of Science and Engineering Research, 5(1), 207-221.

[20] Hossain, M. S., Biswas, B., & Rahaman, M. M. (2023). THE ROLE OF ARTIFICIAL INTELLIGENCE IN ENHANCING CYBERSECURITY DEFENSE MECHANISMS. International Journal Of Engineering Technology Research & Management (IJETRM), 7(07), 157-166.

[21] Biswas, B., Chaudhury, T. H., & Mohammad, S. N. A Fast and Scalable Recommender System using Collaborative Filtering Technique in Python Scikit-learn.

[22]Ansari, A. A., Sazu, A. H., Saleem, M. T., & Chohan, M. A. (2024). FinTech and the liquidity–resilience paradox in stock markets: A critical review. Center for Management Science Research, 2(3), 438–452.