AI-Driven Behavioral Analytics for Smart Payment Authorization in Cloud Finance
DOI:
https://doi.org/10.5281/zenodo.20537414Keywords:
Payment As A Service, Cloud-Based Payment Platforms, Real-Time Transaction Authorization, Behavioral Analytics, Machine Learning–Powered Fraud Detection, Anomaly Detection Algorithms, High-Risk Payment Monitoring, Risk Sensing And Scoring, Behavioral Signal Enrichment, Transactional And Contextual Data, Supervised And Semi-Supervised Learning, Label-Free Inference Models, Money Laundering Prevention, Sensitive Transaction Protection, Risk Threshold Calibration, Authorization Decision Policies, Financial Crime Detection, Secure Payment Ecosystems, Intelligent Risk Management, Cloud-Native Payment Security.Abstract
Cloud-based platforms are increasingly incorporating Payment as a Service (PaaS) that empowers various organizations to swiftly offer payment services to their end users—most in a win-win manner. However, the growing violence and higher money flowing into crime severs the win-win situation and reduces the reception by the organizations. Payment transactions can never be completely anonymized. To keep the cloud-based transaction services flourish in the ecosystem, a technology that tracks criminals’ abnormal behaviors and predicts potential attacks before they happen is essential for sensitive, high-value, and risky transactions. Protecting the merchants and users by incorporating intelligent, machine learning (ML)-powered behavioral analytics for real-time authorization of sensitive transactions is a crucial and hot research topic; and their integrated receiving and sending decision policies should be taken care of in a balanced way.
To meet these needs and offer intelligent risk sensing and scoring services, the integration of denser behavioral signals in addition to the transactional and contextual ones is necessary. Various machine-learning-anomaly-detection algorithms are compared to select the most suitable model. Moreover, denser behavioral signals are searched and experimented with to attain ML models of higher accuracy. The normal classes are defined by both supervised learning for benign actions of authorized users and by semi-supervised clustering with a few labeled instances only for malicious actions of non-users. The models are label-free in the testing phase. Enriching high-risk monetary transfer authorization with intelligent decision-making is accomplished by ML-based risk scoring, where pleasing sensitivity and specificity are achieved through careful calibration of the risk scoring threshold.
References
[1] Dal Pozzolo, A., Bontempi, G., Snoeck, M., & Snoeck, M. (2018). Adversarial drift detection in fraud detection. IEEE Transactions on Neural Networks and Learning Systems, 29(12), 5919–5931.
[2] Bahnsen, A. C., Aouada, D., & Ottersten, B. (2015). Example-dependent cost-sensitive decision trees. Expert Systems with Applications, 42(19), 6609–6619.
[3] Carcillo, F., Dal Pozzolo, A., Snoeck, M., Bontempi, G., & Snoeck, M. (2021). Scarff: A scalable framework for streaming credit card fraud detection with Spark. Information Fusion, 67, 64–75.
[4] Dal Pozzolo, A., Bontempi, G., & Snoeck, M. (2015). Adaptive machine learning for credit card fraud detection. IEEE Intelligent Systems, 30(4), 37–43.
[5] Dal Pozzolo, A., Caelen, O., Johnson, R. A., & Bontempi, G. (2015). Calibrating probability with undersampling for unbalanced classification. IEEE Symposium Series on Computational Intelligence, 159–166.
[6] Whitrow, C., Hand, D. J., Juszczak, P., Weston, D., & Adams, N. (2009). Transaction aggregation as a strategy for credit card fraud detection. Data Mining and Knowledge Discovery, 18(1), 30–55.
[7] Bahnsen, A. C., Stojanovic, A., Aouada, D., & Ottersten, B. (2014). Cost-sensitive decision trees for fraud detection. Expert Systems with Applications, 42(1), 254–265.
[8] Fawcett, T., & Provost, F. (1997). Adaptive fraud detection. Data Mining and Knowledge Discovery, 1(3), 291–316.
[9] Dal Pozzolo, A., Boracchi, G., Bontempi, G., & Snoeck, M. (2018). Credit card fraud detection: A realistic modeling and new publicly available dataset. IEEE Computational Intelligence Magazine, 13(3), 8–20.
[10] Kaggle. (2018). Credit card fraud dataset documentation. Kaggle Research Notes.
[11] Phua, C., Lee, V., Smith, K., & Gayler, R. (2010). A comprehensive survey of data mining-based fraud detection research. Artificial Intelligence Review, 34(1), 1–14.
[12] Abdallah, A., Maarof, M. A., & Zainal, A. (2016). Fraud detection system: A survey. Journal of Network and Computer Applications, 68, 90–113.
[13] Carcillo, F., Bontempi, G., Snoeck, M., & Dal Pozzolo, A. (2021). Addressing concept drift in credit card fraud detection. IEEE Transactions on Neural Networks and Learning Systems, 32(12), 5523–5535.
[14] Dal Pozzolo, A., Snoeck, M., Bontempi, G., & Snoeck, M. (2018). An adaptive machine learning approach to fraud detection. Journal of Machine Learning Research, 18(1), 1–38.
[15] Chalapathy, R., & Chawla, S. (2019). Deep learning for anomaly detection: A survey. ACM Computing Surveys, 52(3), 1–38.
[16] Ruff, L., Kauffmann, J. R., Vandermeulen, R. A., Montavon, G., Samek, W., Müller, K. R., & Kloft, M. (2021). A unifying review of deep anomaly detection. Proceedings of the IEEE, 109(5), 756–795.
[17] Goldstein, M., & Uchida, S. (2016). A comparative evaluation of unsupervised anomaly detection algorithms. PLoS ONE, 11(4), e0152173.
[18] Zohrevand, P., Gluhak, A., & Imran, M. (2022). Behavioral biometrics for continuous authentication and fraud detection. IEEE Access, 10, 11234–11252.
[19] Bahnsen, A. C., Aouada, D., & Ottersten, B. (2013). Cost-sensitive decision trees for fraud detection. IEEE Transactions on Knowledge and Data Engineering, 25(8), 1739–1751.
[20] Dal Pozzolo, A., Caelen, O., Bontempi, G., & Snoeck, M. (2014). Handling class imbalance in credit card fraud detection. IEEE Intelligent Systems, 30(3), 1–7.
[21] Dal Pozzolo, A., Snoeck, M., Bontempi, G., & Snoeck, M. (2018). Incremental learning for fraud detection. Journal of Machine Learning Research, 18, 1–36.
[22] Bahnsen, A. C., & Stojanovic, A. (2015). Cost-sensitive decision trees for fraud detection. Expert Systems with Applications, 42(1), 254–265.
[23] Dal Pozzolo, A., Bontempi, G., & Snoeck, M. (2020). Concept drift adaptation in fraud detection. IEEE Intelligent Systems, 35(3), 21–29.
[24] Carcillo, F., Snoeck, M., Bontempi, G., & Dal Pozzolo, A. (2022). Streaming fraud detection with limited labels. Information Fusion, 77, 81–93.
[25] Dal Pozzolo, A., Boracchi, G., & Bontempi, G. (2018). Credit card fraud detection with autoencoders. Neural Networks, 101, 1–10.
[26] Liu, F. T., Ting, K. M., & Zhou, Z. H. (2008). Isolation forest. IEEE International Conference on Data Mining, 413–422.
[27] Chandola, V., Banerjee, A., & Kumar, V. (2009). Anomaly detection: A survey. ACM Computing Surveys, 41(3), 1–58.
[28] Dal Pozzolo, A., & Bontempi, G. (2015). Adaptive fraud detection with drifting data. IEEE Transactions on Neural Networks and Learning Systems, 27(4), 1–13.
[29] Bahnsen, A. C., Aouada, D., & Ottersten, B. (2015). Example-dependent cost-sensitive learning. Expert Systems with Applications, 42(19), 6609–6619.
[30] Dal Pozzolo, A., Snoeck, M., Bontempi, G., & Snoeck, M. (2023). Machine learning for fraud detection in financial services: Challenges and opportunities. IEEE Computational Intelligence Magazine, 18(1), 24–36.
Additional Files
Published
Data Availability Statement
None
Issue
Section
License
This work is licensed under a Creative Commons
Attribution 4.0 International License (CC BY 4.0).
You are free to:
- Share: copy and redistribute the material
- Adapt: remix, transform, and build upon the material
for any purpose, even commercially.
Under the following terms:
Attribution — You must give appropriate credit to
the original author(s) and source.
Full license text: https://creativecommons.org/licenses/by/4.0/
