Advancements of Credit Card Fraud Detection Based on Federated Learning

Hongwei Wang

2024

Abstract

Credit card fraud is an increasing concern as digital payment systems and e-commerce continue to expand. The purpose of this research is to investigate the use of federated learning (FL) to identify credit card fraud while maintaining the privacy of data across financial institutions. Unlike traditional centralized models, federated learning can let to train models collaboratively without disclosing raw data. This study reviews critical federated learning approaches, such as vertical and horizontal FL, and highlights their effectiveness in addressing privacy concerns, data heterogeneity, and class imbalance. Integrating advanced techniques like expert systems, explainable Artificial Intelligence (AI) tools and hybrid resampling methods enhances AI models' detection accuracy and transparency. Federated learning offers significant advantages for safeguarding financial transactions by improving real-time fraud detection and adaptive learning. The study also identifies challenges like model scalability, communication overhead, and security threats like inference attacks. Nonetheless, the potential for federated learning to transform credit card fraud detection by combining privacy-preserving, interpretable, and scalable solutions is considerable, positioning it as a critical technology in the prevent financial fraud.

Paper Citation

in Harvard Style

Wang H. (2024). Advancements of Credit Card Fraud Detection Based on Federated Learning. In Proceedings of the 2nd International Conference on Data Analysis and Machine Learning - Volume 1: DAML; ISBN 978-989-758-754-2, SciTePress, pages 511-515. DOI: 10.5220/0013527400004619

in Bibtex Style

@conference{daml24,
author={Hongwei Wang},
title={Advancements of Credit Card Fraud Detection Based on Federated Learning},
booktitle={Proceedings of the 2nd International Conference on Data Analysis and Machine Learning - Volume 1: DAML},
year={2024},
pages={511-515},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0013527400004619},
isbn={978-989-758-754-2},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 2nd International Conference on Data Analysis and Machine Learning - Volume 1: DAML
TI - Advancements of Credit Card Fraud Detection Based on Federated Learning
SN - 978-989-758-754-2
AU - Wang H.
PY - 2024
SP - 511
EP - 515
DO - 10.5220/0013527400004619
PB - SciTePress