Credit Card Fraud Detection with Advanced Hybrid Deep Learning Algorithm

Abstract

Progress in the rise of online payments systems has made shopping easier. However, it also increased credit card fraud, with losses expected to exceed $362 billion by 2027. Traditional machine learning algorithms often fail to detect fraud due to complex patterns and unbalanced data. This study uses the IEEE-CIS dataset over which machine learning (ML) and deep learning (DL) algorithms such as Decision Trees, XGBoost, LightGBM, and an advanced Hybrid Deep Learning Model were applied for credit card fraud detection. The proposed model gave the best results, with an ROC AUC of 0.9709 and F1-score of 0.9103. The paper suggests that combining deep learning with explainable AI can improve fraud detection, and future research will aim to make models more transparent and efficient.

Introduction

The rapid growth of online payments has increased convenience but also escalated credit card fraud, with losses projected to exceed $362 billion by 2027. Traditional rule-based detection struggles with evolving fraud patterns, imbalanced datasets, and real-time requirements. Machine learning (ML) and deep learning (DL) approaches have shown promise, but challenges remain in reducing false positives, handling data imbalance, and ensuring model interpretability.

Literature Insights:

• Common challenges include class imbalance, feature redundancy, and evolving fraud techniques.

• Techniques like SMOTE, K-SMOTEENN, and federated learning help balance datasets and preserve privacy.

• Ensemble methods (stacking, XGBoost, CatBoost, LightGBM) and hybrid DL architectures (CNN + LSTM + Transformers) improve detection accuracy and robustness.

• Explainable AI (XAI) techniques like SHAP and LIME enhance interpretability.

• Integrating AI with blockchain and real-time monitoring provides scalable, tamper-proof solutions.

Research Methodology:

• Dataset: IEEE-CIS Fraud Detection dataset (~590k transactions, 3.5% fraud).

• Preprocessing: Missing value imputation, encoding, resampling, and top-feature selection for training.

• Models Developed: Decision Tree (baseline), LightGBM, CatBoost, XGBoost, and an enhanced neural network with advanced layers, multi-head attention, residual and squeeze-excite blocks.

• Training & Evaluation: 80/20 train-test split, feature scaling, specialized loss functions (Focal, Class-Balanced, Binary Cross-Entropy), and threshold optimization.

Results:

• LightGBM: ROC AUC 0.9491

• CatBoost: ROC AUC 0.9407

• XGBoost: ROC AUC 0.9600

• Enhanced Neural Network: ROC AUC 0.9709, F1-score 0.9103, precision/recall ~0.91, accuracy 91%

• Advanced feature engineering and specialized loss functions improved performance by 3–5%.

Conclusion

In this study, we presented a comprehensive methodology that effectively integrates data preprocessing, feature selection, class balancing, and advanced model development for fraud detection. Among all models evaluated, the enhanced neural network outperformed traditional approaches, achieving the highest ROC AUC score of 0.9709 along with strong precision, recall, and F1-score. The application of advanced feature engineering and specialized loss functions contributed signifi- cantly to this performance increase. Despite promising results, challenges such as feature anonymity, potential oversampling noise, and computational limitations were observed. Future work can focus on exploring more interpretable models, integrating attention mechanisms for better feature attribution, and deploying explainable AI (XAI) frameworks to improve transparency.

References

[1] R. K. Gupta, A. Hassan, S. K. Majhi, N. Parveen, A. T. Zamani, R. Anitha, B. Ojha, A. K. Singh, and D. Muduli, “Enhanced framework for credit card fraud detection using robust feature selection and a stacking ensemble model approach,” Results in Engineering, vol. 26, p. 105084, Apr. 2025. [2] Y. Chen, C. Zhao, Y. Xu, and C. Nie, “Year-over-year developments in financial fraud detection via deep learning: A systematic literature review,” Jan. 2025. [3] W. Mohamedhen, M. Charfeddine, and Y. Hadj Kacem, “Enhanced credit card fraud detection using federated learning, lstm models, and the smote technique,” in Proc. 17th Int. Conf. Agents Artif. Intell. (ICAART), vol. 3, 2025, pp. 368–375. [4] M. Gostkowski, A. Krasnode?bski, and A. Niedzió?ka, “Credit card fraud detection using machine learning techniques,” European Research Studies Journal, vol. 27, no. 2, pp. 571–585, 2024. [5] F. Almalki and M. Masud, “Financial fraud detection using explainable ai and stacking ensemble methods,” May 2025. [6] Y. Wu, L. Wang, H. Li, and J. Liu, “A deep learning method of credit card fraud detection based on continuous-coupled neural networks,” Mathematics, vol. 13, no. 5, p. 819, Feb. 2025. [7] R. V. Karunya, S. Ganesh, D. Muralidharan, G. R. Brindha, and M. Thiruvengadam, “Credit card fraud data analysis and prediction using machine learning algorithms,” Security and Privacy, vol. 8, no. 3, [8] p. e413, Apr. 2025. [9] S. F. Farabi, M. Prabha, M. Alam, M. Z. Hossan, M. Arif, M. R. Islam, A. Uddin, M. Bhuiyan, and M. Z. A. Biswas, “Enhancing credit card fraud detection: A comprehensive study of machine learning algorithms and performance evaluation,” Journal of Business and Management Studies, vol. 6, no. 4, pp. 1–10, 2024. [10] B. Fetaji, M. Fetaji, A. Hasan, S. Rexhepi, and G. Armenski, “Fraud-x: An integrated ai, blockchain, and cybersecurity framework with early warning systems for mitigating online financial fraud: A case study from north macedonia,” IEEE Access, vol. 13, pp. 35 472–35 485, Mar. 2025. [11] A. El Bazi, M. Chrayah, N. Aknin, and A. Bouzidi, “Enhancing credit card fraud detection using a stacking model approach and hyperparameter optimization,” International Journal of Advanced Computer Science and Applications (IJACSA), vol. 15, no. 10, pp. 1080–1087, 2024. [12] N. Damanik and C.-M. Liu, “Advanced fraud detection: Leveraging k-smoteenn and stacking ensemble to tackle data imbalance and extract insights,” IEEE Access, vol. 13, pp. 35 280–35 290, Jan. 2025. [13] E. Ileberi and Y. Sun, “A hybrid deep learning ensemble model for credit card fraud detection,” IEEE Access, vol. 12, pp. 175 829–175 842, Nov. 2024.

Copyright

Copyright © 2026 Sumedha Arya, Nirmal Gaud. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.