Efficient Confidential Credit Risk Modeling Using Homomorphic Encryption and Approximation-Aware Deep Learning for Secure Financial Data Analytics

Authors

  • Kacper Daniec Department of Computer Science, Colorado State University, Fort Collins, CO, USA.
  • Laureano Curci School of Electrical Engineering and Computer Science, Oregon State University, Corvallis, OR, USA.
  • Ubika Yogi Department of Computer Science and Engineering, University at Buffalo, Buffalo, NY, USA.

Keywords:

homomorphic encryption; credit risk modeling; approximation-aware deep learning; CKKS; calibration; secure inference

Abstract

Credit risk modeling depends on sensitive financial and behavioral data, which creates a tension between statistical precision and confidentiality when models are trained or evaluated across organizational boundaries. Homomorphic encryption supports computation on encrypted inputs, but secure neural inference for tabular credit data is constrained by multiplicative depth, noise growth, activation approximation error, and the need for calibrated probability-of-default estimates. This article develops an approximation-aware confidential credit risk modeling framework that combines CKKS approximate homomorphic encryption with low-depth polynomial neural networks, calibration-aware training, and depth-conscious architecture selection. The proposed formulation treats encrypted inference as a constrained approximation problem in which activation polynomials, plaintext growth bounds, modulus-chain sizing, and decision thresholds are selected jointly rather than sequentially. Credit-specific requirements, including probability calibration, rank preservation, monotonic risk factors, auditability, and cost-sensitive default decisions, are integrated into the model design and evaluation logic. The analysis clarifies how approximation error propagates through affine layers, why minimax activation fitting is preferable to unconstrained regression of nonlinearities, and how ciphertext packing and parameter folding affect latency and noise consumption. It also identifies failure modes that are specific to encrypted credit analytics, including interval violation, overconfident calibrated scores, noise-budget exhaustion, and mismatches between encrypted-score computation and downstream decision rules. The resulting methodological framework provides a defensible basis for secure financial data analytics without treating cryptographic efficiency and credit-model validity as separate objectives.

References

1. Gentry, C. (2009). Fully homomorphic encryption using ideal lattices. In Proceedings of the 41st Annual ACM Symposium on Theory of Computing (pp. 169–178). ACM.

2. Cheon, J. H., Kim, A., Kim, M., & Song, Y. (2017). Homomorphic encryption for arithmetic of approximate numbers. In Advances in Cryptology – ASIACRYPT 2017 (pp. 409–437). Springer.

3. Gilad-Bachrach, R., Dowlin, N., Laine, K., Lauter, K., Naehrig, M., & Wernsing, J. (2016). CryptoNets: Applying neural networks to encrypted data with high throughput and accuracy. In Proceedings of the 33rd International Conference on Machine Learning (pp. 201–210). PMLR.

4. Bost, R., Popa, R. A., Tu, S., & Goldwasser, S. (2015). Machine learning classification over encrypted data. In Proceedings of the 2015 Network and Distributed System Security Symposium. Internet Society.

5. Mohassel, P., & Zhang, Y. (2017). SecureML: A system for scalable privacy-preserving machine learning. In 2017 IEEE Symposium on Security and Privacy (pp. 19–38). IEEE.

6. Juvekar, C., Vaikuntanathan, V., & Chandrakasan, A. (2018). GAZELLE: A low latency framework for secure neural network inference. In 27th USENIX Security Symposium (pp. 1651–1669). USENIX Association.

7. Mishra, P., Lehmkuhl, R., Srinivasan, A., Zheng, W., & Popa, R. A. (2020). Delphi: A cryptographic inference service for neural networks. In 29th USENIX Security Symposium (pp. 2505–2522). USENIX Association.

8. Lou, Q., & Jiang, L. (2019). SHE: A fast and accurate deep neural network for encrypted data. In Advances in Neural Information Processing Systems 32. Curran Associates.

9. Boulemtafes, A., Derhab, A., & Challal, Y. (2020). A review of privacy-preserving techniques for deep learning. Neurocomputing, 384, 21–45.

10. Al Badawi, A., Bates, J., Bergamaschi, F., Cousins, D. B., Ead, S., Eads, C., ... & Polyakov, Y. (2022). OpenFHE: Open-source fully homomorphic encryption library. In Proceedings of the 10th Workshop on Encrypted Computing and Applied Homomorphic Cryptography. ACM.

11. Hand, D. J., & Henley, W. E. (1997). Statistical classification methods in consumer credit scoring: A review. Journal of the Royal Statistical Society: Series A (Statistics in Society), 160(3), 523–541.

12. Thomas, L. C., Edelman, D. B., & Crook, J. N. (2002). Credit scoring and its applications. SIAM.

13. Lessmann, S., Baesens, B., Seow, H.-V., & Thomas, L. C. (2015). Benchmarking state-of-the-art classification algorithms for credit scoring: An update of research. European Journal of Operational Research, 247(1), 124–136.

14. Khandani, A. E., Kim, A. J., & Lo, A. W. (2010). Consumer credit-risk models via machine-learning algorithms. Journal of Banking & Finance, 34(11), 2767–2787.

15. Barboza, F., Kimura, H., & Altman, E. (2017). Machine learning models and bankruptcy prediction. Expert Systems with Applications, 83, 405–417.

16. Peng, H., Huang, S., Zhou, T., Luo, Y., Wang, C., Wang, Z., ... & Ding, C. (2023, October). Autorep: Automatic relu replacement for fast private network inference. In 2023 IEEE/CVF International Conference on Computer Vision (ICCV) (pp. 5155-5165). IEEE.

17. Hesamifard, E., Takabi, H., & Ghasemi, M. (2017). CryptoDL: Deep neural networks over encrypted data. arXiv preprint arXiv:1711.05189.

18. Kim, M., Song, Y., Wang, S., Xia, Y., & Jiang, X. (2018). Secure logistic regression based on homomorphic encryption: Design and evaluation. JMIR Medical Informatics, 6(2), e19.

19. Benaissa, A., Retiat, B., Cebere, B., & Belfodil, A. E. (2021). TenSEAL: A library for encrypted tensor operations using homomorphic encryption. arXiv preprint arXiv:2104.03152.

20. Naehrig, M., Lauter, K., & Vaikuntanathan, V. (2011). Can homomorphic encryption be practical? In Proceedings of the 3rd ACM Workshop on Cloud Computing Security Workshop (pp. 113–124). ACM.

21. Niculescu-Mizil, A., & Caruana, R. (2005). Predicting good probabilities with supervised learning. In Proceedings of the 22nd International Conference on Machine Learning (pp. 625–632). ACM.

22. Acar, A., Aksu, H., Uluagac, A. S., & Conti, M. (2018). A survey on homomorphic encryption schemes: Theory and implementation. ACM Computing Surveys, 51(4), Article 79.

Downloads

Published

2026-09-14

How to Cite

Kacper Daniec, Laureano Curci, & Ubika Yogi. (2026). Efficient Confidential Credit Risk Modeling Using Homomorphic Encryption and Approximation-Aware Deep Learning for Secure Financial Data Analytics. Global Financial Analytics Research Review, 3(4). Retrieved from https://gfarr.org/index.php/home/article/view/174