Adversarial Robustness in Financial Machine Learning: Defenses, Economic Impact, and Governance Evidence
We evaluate adversarial robustness in tabular machine learning models used in financial decision making. Using credit scoring and fraud detection data, we apply gradient based attacks and measure impacts on discrimination, calibration, and financial risk metrics. Results show notable performance degradation under small perturbations and partial recovery through adversarial training.
Paper
References (11)
07Secure credit risk assessment with machine learning2021 · Management Science
08Robust decision-making under distributional shift in tabular domains2020 · Advances in Neural Information Processing Systems (NeurIPS)
09Adversarial robustness in fraud detection systems2020 · ACM Conference on Computer and Communications Security
10Defense Strategies: Evaluation of both robust PGD adversarial training and efficient, lightweight defenses such as noise and gradient regularization techniques
11Adversarial examples in tabular machine learningarXiv