Banking systems are increasingly exposed to fraud in real-time digital transactions. Rule-based and single-model detection systems are reported to struggle with evolving fraud patterns, high false-positive rates and severe class imbalance. Hybrid deep learning (DL) approaches, such as convolutional neural networks (CNN) combined with long short-term memory (LSTM) or bidirectional LSTM (BiLSTM) and attention mechanisms, report strong offline performance, but they are seldom designed or evaluated for real-time deployment. This paper surveys machine learning (ML) and DL techniques for transaction fraud detection and analyses their advantages and limitations. From this survey, it identifies a gap between high-performing offline models and the requirements of real-time, scalable and explainable operation. In response, it proposes a conceptual architecture that combines a PyTorch-based CNN-BiLSTM engine, SHapley Additive exPlanations (SHAP) for explainability, and a dual-agent LangChain framework (a SOC Assistant and a User Verification Agent) for risk-based verification and response. The proposed architecture has not yet been experimentally validated. This paper reports no experimental results, and the evaluation plan is described as future work.
- G. R. Jainish and P. Alwin Infant, “Attention layer integrated BiLSTM for financial fraud prediction,” Multimedia Tools and Applications, vol. 83, no. 34, pp. 80613–80629, 2024.
- N. Madhusudhana Reddy, K. A. Sharada, D. Pilli, R. N. Paranthaman, K. S. Reddy, and A. Chauhan, “CNN-bidirectional LSTM based approach for financial fraud detection and prevention system,” in Proc. Int. Conf. Sustainable Computing and Smart Systems (ICSCSS), 2023.
- A. R. Khalid et al., “Enhancing credit card fraud detection: An ensemble machine learning approach,” Big Data and Cognitive Computing, vol. 8, no. 1, 2024.
- Bagga et al., “Ensemble learning pipeline,” 2020.
- S. Carbo-Valverde et al., “A machine learning approach to the digitalization of bank customers: Evidence from random and causal forests,” 2020.
- I. D. Mienye and N. Jere, “Deep learning for credit card fraud detection: A review of algorithms, challenges, and solutions,” IEEE Access, vol. 12, pp. 96893–96910, 2024, doi: 10.1109/ACCESS.2024.3426955.
- E. Btoush et al., “A systematic review of literature on credit card cyber fraud detection using machine and deep learning,” PeerJ Computer Science, 2023.
- Sharifani and Amini, “Machine learning and deep learning literature review,” 2023.
- Agarwal et al., “Hybrid CNN-BiLSTM-attention,” 2021.
- Alarfaj et al., “State-of-the-art ML/DL algorithms,” 2022.
- Bandyopadhyay et al., “Ensemble and clustering-based anomaly detection,” 2021.
- Berhane et al., “Hybrid CNN-SVM,” 2023.
- Nama, Obaid, and Alrammahi, “Deep learning plus SVM hybrid,” 2023.