A Credit Card Fraud Detection Algorithm Based on SDT and Federated Learning

被引:0
作者
Tang, Yuxuan [1 ]
Liu, Zhanjun [2 ]
机构
[1] Southwestern Univ Finance & Econ, Sch Accounting, Chengdu 611130, Sichuan, Peoples R China
[2] Chongqing Univ Posts & Telecommun, Sch Commun & Informat Engn, Chongqing 400065, Peoples R China
来源
IEEE ACCESS | 2024年 / 12卷
关键词
Data models; Fraud; Credit cards; Transformers; Predictive models; Feature extraction; Federated learning; Adaptation models; Accuracy; Distributed databases; Attention mechanisms; deep learning; federated learning; fraud detection; transformer;
D O I
10.1109/ACCESS.2024.3491175
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
With the rise of digital payment methods and the growth in financial transactions, the issue of credit card fraud has become increasingly severe. Traditional fraud detection methods are currently facing challenges such as poor model performance, difficulty in obtaining accurate results, and limitations in distributed deployment. These challenges stem from constantly evolving fraud strategies, higher volumes of transactions, and the complexity of the financial environment. This study proposes a credit card fraud detection algorithm based on Structured Data Transformer (SDT) and federated learning, which leverages the advanced capabilities of the Transformer model in deep learning. First, we organize credit card data into sequences and introduce a special, learnable token at the beginning of each sequence for classification purposes. Thanks to the attention mechanism of the Transformer, the model can automatically highlight important features in the data, significantly improving the accuracy of fraud detection. Second, addressing the complex financial environment and concerns about financial data privacy, we introduce a federated learning architecture to deploy the SDT model across different banks in a distributed manner. Momentum updates are used for model parameter updates during training, which enhance model performance and ensure data privacy between banks. Lastly, we conducted experimental validation on two financial datasets of different scales. The results on Dataset 1 and Dataset 2 show that our proposed SDT model surpasses traditional detection methods in terms of AUC-PR values (0.882, 0.816) and AUC-ROC values (0.982, 0.994). By integrating federated learning and deploying and testing the two datasets in a distributed environment, the AUC-PR values (0.884, 0.892) and AUC-ROC values (0.963, 0.998) can be further improved.
引用
收藏
页码:182547 / 182560
页数:14
相关论文
共 33 条
  • [1] Enhanced Credit Card Fraud Detection Model Using Machine Learning
    Alfaiz, Noor Saleh
    Fati, Suliman Mohamed
    [J]. ELECTRONICS, 2022, 11 (04)
  • [2] Facilitating User Authorization from Imbalanced Data Logs of Credit Cards Using Artificial Intelligence
    Arora, Vinay
    Leekha, Rohan Singh
    Lee, Kyungroul
    Kataria, Aman
    [J]. MOBILE INFORMATION SYSTEMS, 2020, 2020
  • [3] Combining unsupervised and supervised learning in credit card fraud detection
    Carcillo, Fabrizio
    Le Borgne, Yann-Ael
    Caelen, Olivier
    Kessaci, Yacine
    Oble, Frederic
    Bontempi, Gianluca
    [J]. INFORMATION SCIENCES, 2021, 557 : 317 - 331
  • [4] SCARFF: A scalable framework for streaming credit card fraud detection with spark
    Carcillo, Fabrizio
    Dal Pozzolo, Andrea
    Le Borgne, Yann-Ael
    Caelen, Olivier
    Mazzer, Yannis
    Bontempi, Gianluca
    [J]. INFORMATION FUSION, 2018, 41 : 182 - 194
  • [5] Dosovitskiy A, 2021, Arxiv, DOI arXiv:2010.11929
  • [6] El hlouli F. Z., 2020, P INT C INT SYST COM, P1, DOI [10.1109/ISCV49265.2020.9204185, DOI 10.1109/ISCV49265.2020.9204185]
  • [7] Ensemble of deep sequential models for credit card fraud detection
    Forough, Javad
    Momtazi, Saeedeh
    [J]. APPLIED SOFT COMPUTING, 2021, 99
  • [8] Greedy function approximation: A gradient boosting machine
    Friedman, JH
    [J]. ANNALS OF STATISTICS, 2001, 29 (05) : 1189 - 1232
  • [9] A Convolutional Neural Network Model for Credit Card Fraud Detection
    Gambo, Muhammad Liman
    Zainal, Anazida
    Kassim, Mohamad Nizam
    [J]. 2022 INTERNATIONAL CONFERENCE ON DATA SCIENCE AND ITS APPLICATIONS (ICODSA), 2022, : 198 - 202
  • [10] Learning Sequential Behavior Representations for Fraud Detection
    Guo, Jia
    Liu, Guannan
    Zuo, Yuan
    Wu, Junjie
    [J]. 2018 IEEE INTERNATIONAL CONFERENCE ON DATA MINING (ICDM), 2018, : 127 - 136