Generative deep learning for class imbalance in credit risk modeling: comparative evaluation of synthetic oversampling methods and an elastic oversampling extension
This paper applies generative deep learning–GAN, VAE, and CVAE–to oversample rare defaults in credit risk modeling, with a posterior binarization procedure introduced to preserve the marginal distribution of dummy features. Using firm- and individual-level default datasets, we compare these methods against SMOTE, SMOTE-Tomek, and ADASYN under multiple classifiers. Empirical results show that synthetic oversampling improves recall and disperses feature importance, while generative oversampling tends to sharpen class boundaries, favor precision, and concentrate feature importance. To balance these complementary strengths, we introduce Elastic Oversampling, a parallel hybrid of generative and synthetic approaches with a tunable mixing ratio for the precision–recall trade-off. Elastic Oversampling emphasizes distinctive minority cases through generative methods while constraining unrealistic samples with synthetic methods, offering a flexible and policy-sensitive framework for class-imbalanced credit risk prediction.
Generative deep learning for class imbalance in credit risk modeling: comparative evaluation of synthetic oversampling methods and an elastic oversampling extension. // International Journal of the Classical Tradition. – 2026. – DOI: 10.1186/s40854-026-00970-5. – URL: https://link.springer.com/article/10.1186/s40854-026-00970-5
(2026). Generative deep learning for class imbalance in credit risk modeling: comparative evaluation of synthetic oversampling methods and an elastic oversampling extension. International Journal of the Classical Tradition. https://doi.org/10.1186/s40854-026-00970-5
“Generative deep learning for class imbalance in credit risk modeling: comparative evaluation of synthetic oversampling methods and an elastic oversampling extension.” International Journal of the Classical Tradition, 2026. doi:10.1186/s40854-026-00970-5.
@article{article33828,
title = {Generative deep learning for class imbalance in credit risk modeling: comparative evaluation of synthetic oversampling methods and an elastic oversampling extension},
journal = {International Journal of the Classical Tradition},
year = {2026},
doi = {10.1186/s40854-026-00970-5},
url = {https://link.springer.com/article/10.1186/s40854-026-00970-5}
}
This paper links climate science with sovereign risk assessment to produce a single forward-looking measure of country-level climate change risk. We combine the Network for Greening the Financial System (NGFS) climate scenarios with a...
This paper assesses the relationship between credit risk and different climate variables under a system of equations tested using seemingly unrelated regression. When considering both total emissions and Scope 1 emissions only, carbon...
An environmental limit, such as a carbon budget, changes the nature of economic growth, but does not preclude it. When atmospheric greenhouse gas levels reach a critical threshold, further economic growth requires a permanently...
Frontier technologies, such as artificial intelligence, when deeply integrated with traditional finance, give rise to financial technology (Fintech), which continuously influences cross-regional capital flows. We examine the development...
Pandemic bonds, as a subtype of catastrophe (CAT) bonds, are designed to transfer public health risks to capital markets, yet they face substantial liquidity risk due to their nascent market structure and limited risk design. Using a...