RAPTOR: RISK-ADAPTIVE SECURITY AND PRIVACY-PRESERVING TRANSACTION PROCESSING FOR LARGE-SCALE MOBILE PAYMENT SYSTEMS
Keywords:
Mobile payments, Fraud detection, Federated learning, Differential privacy, Secure aggregation, Risk-based authentication, Data minimizationAbstract
Mobile payment platforms must decide within milliseconds whether to approve, challenge or decline each transaction, while the most informative fraud signals are personal behavioural data that are fragmented across payment service providers (PSPs) and protected by data-protection law. This paper presents RAPTOR, a risk-adaptive and privacy-preserving transaction-processing framework that allocates both privacy exposure and authentication friction in proportion to the money at risk. RAPTOR (i) trains a two-stage risk model across PSPs with cross-silo federated learning, record-level differentially private SGD and pairwise-masking secure aggregation, with Rényi-DP accounting over both stages; (ii) scores every request with a Stage-1 model that uses only non-personal context and reads the customer profile through a Stage-2 model only when the prior-smoothed expected loss of the request exceeds a coverage-calibrated threshold; and (iii) maps the final calibrated risk to frictionless approval, step-up strong customer authentication or decline with an expected-loss rule that provably maximizes the number of exemptions under a fraud-loss budget. Experiments on two public payment benchmarks with 0.59 M and 1.75 M transactions distributed over five PSPs show that federated training stays within 0.012 AP of a centralized model, that the escalation gate routes 90.9% of the fraud value to profile-based verification while reading customer profiles for only 14.8% of the transactions, and that the expected-loss policy exempts 75.1% of the transactions at the reference loss budget, compared with 70.1% for a risk threshold and 35.2% for a low-value exemption rule. The online path sustains 3.25 M transactions per second on a single CPU core.References
[1] European Commission. Commission Delegated Regulation (EU) 2018/389 of 27 November 2017 supplementing Directive (EU) 2015/2366 with regard to regulatory technical standards for strong customer authentication and common and secure open standards of communication. Official Journal of the European Union, 2018, L 69: 23-43.
[2] European Parliament and Council. Regulation (EU) 2016/679 (General Data Protection Regulation). Official Journal of the European Union, 2016, L 119: 1-88.
[3] Dal Pozzolo A, Boracchi G, Caelen O, et al. Credit card fraud detection: A realistic modeling and a novel learning strategy. IEEE Transactions on Neural Networks and Learning Systems, 2018, 29(8): 3784-3797.
[4] Correa Bahnsen A, Aouada D, Stojanovic A, et al. Feature engineering strategies for credit card fraud detection. Expert Systems with Applications, 2016, 51: 134-142.
[5] Jurgovsky J, Granitzer M, Ziegler K, et al. Sequence classification for credit-card fraud detection. Expert Systems with Applications, 2018, 100: 234-245.
[6] McMahan B, Moore E, Ramage D, et al. Communication-efficient learning of deep networks from decentralized data. In: Proceedings of the 20th International Conference on Artificial Intelligence and Statistics (AISTATS), 2017: 1273-1282.
[7] Yang Q, Liu Y, Chen T, et al. Federated machine learning: Concept and applications. ACM Transactions on Intelligent Systems and Technology, 2019, 10(2): 12:1-12:19.
[8] Kairouz P, McMahan H B, et al. Advances and open problems in federated learning. Foundations and Trends in Machine Learning, 2021, 14(1-2): 1-210.
[9] Dwork C, Roth A. The algorithmic foundations of differential privacy. Foundations and Trends in Theoretical Computer Science, 2014, 9(3-4): 211-407.
[10] Abadi M, Chu A, Goodfellow I, et al. Deep learning with differential privacy. In: Proceedings of the ACM SIGSAC Conference on Computer and Communications Security (CCS), 2016: 308-318.
[11] Yang W, Zhang Y, Ye K, et al. FFD: A federated learning based method for credit card fraud detection. In: Proceedings of the International Conference on Big Data (BigData 2019), LNCS 11514, Springer, 2019: 18-32.
[12] Zheng W, Yan L, Gou C, et al. Federated meta-learning for fraudulent credit card detection. In: Proceedings of the 29th International Joint Conference on Artificial Intelligence (IJCAI), 2020: 4654-4660.
[13] Byrd D, Polychroniadou A. Differentially private secure multi-party computation for federated learning in financial applications. In: Proceedings of the 1st ACM International Conference on AI in Finance (ICAIF), New York, NY, USA, 2020.
[14] Freeman D, Jain S, Dürmuth M, et al. Who are you? A statistical approach to measuring user authenticity. In: Proceedings of the Network and Distributed System Security Symposium (NDSS), 2016.
[15] Wiefling S, Lo Iacono L, Dürmuth M. Is this really you? An empirical study on risk-based authentication applied in the wild. In: Proceedings of the 34th IFIP TC 11 International Conference on ICT Systems Security and Privacy Protection (SEC), IFIP AICT 562, 2019: 134-148.
[16] Chen T, Guestrin C. XGBoost: A scalable tree boosting system. In: Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 2016: 785-794.
[17] Ke G, Meng Q, Finley T, et al. LightGBM: A highly efficient gradient boosting decision tree. In: Advances in Neural Information Processing Systems (NIPS), 2017: 3146-3154.
[18] Lopez-Rojas E A, Axelsson S. BankSim: A bank payments simulator for fraud detection research. In: Proceedings of the 26th European Modeling and Simulation Symposium (EMSS), Bordeaux, France, 2014: 144-152.
[19] Lopez-Rojas E A, Elmir A, Axelsson S. PaySim: A financial mobile money simulator for fraud detection. In: Proceedings of the 28th European Modeling and Simulation Symposium (EMSS), Larnaca, Cyprus, 2016: 249-255.
[20] Le Borgne Y A, Siblini W, Lebichot B, et al. Reproducible Machine Learning for Credit Card Fraud Detection - Practical Handbook. Brussels, Belgium: Université Libre de Bruxelles, 2022.
[21] McMahan H B, Ramage D, Talwar K, et al. Learning differentially private recurrent language models. In: Proceedings of the International Conference on Learning Representations (ICLR), 2018.
[22] Mironov I. Rényi differential privacy. In: Proceedings of the IEEE 30th Computer Security Foundations Symposium (CSF), 2017: 263-275.
[23] Mironov I, Talwar K, Zhang L. Rényi differential privacy of the sampled Gaussian mechanism. arXiv:1908.10530, 2019.
[24] Bonawitz K, Ivanov V, Kreuter B, et al. Practical secure aggregation for privacy-preserving machine learning. In: Proceedings of the ACM SIGSAC Conference on Computer and Communications Security (CCS), 2017: 1175-1191.
[25] Elkan C. The foundations of cost-sensitive learning. In: Proceedings of the 17th International Joint Conference on Artificial Intelligence (IJCAI), 2001: 973-978.
[26] Zadrozny B, Elkan C. Transforming classifier scores into accurate multiclass probability estimates. In: Proceedings of the 8th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 2002: 694-699.
[27] Xu Z, Kusner M J, Weinberger K Q, et al. Classifier cascades and trees for minimizing feature evaluation cost. Journal of Machine Learning Research, 2014, 15: 2113-2144.