FinGraphFL: Financial Graph-Based Federated Learning for Enhanced Credit Card Fraud Detection

Xia, Z; Saha, SC

FinGraphFL: Financial Graph-Based Federated Learning for Enhanced Credit Card Fraud Detection

Xia, Z Saha, SC

Permalink

Publisher:: MDPI
Publication Type:: Journal Article
Citation:: Mathematics, 2025, 13, (9)
Issue Date:: 2025-05-01

Open Access

Copyright Clearance Process

Recently Added
In Progress
Open Access

This item is open access.

Adobe PDF

Download Published versionAdobe PDF (811.72 kB)

View on publisher's site

View statistics

Full metadata record

Field	Value	Language
dc.contributor.author	Xia, Z
dc.contributor.author	Saha, SC
dc.date.accessioned	2025-10-09T04:33:37Z
dc.date.available	2025-10-09T04:33:37Z
dc.date.issued	2025-05-01
dc.identifier.citation	Mathematics, 2025, 13, (9)
dc.identifier.issn	2227-7390
dc.identifier.issn	2227-7390
dc.identifier.uri	http://hdl.handle.net/10453/190331
dc.description.abstract	In the field of credit card fraud detection, traditional methods often struggle due to their reliance on complex manual feature engineering or their inability to adapt to rapidly changing fraud patterns. This paper introduces an innovative approach called FinGraphFL, which merges graph-based learning with the principles of federated learning and improves security through differential privacy. FinGraphFL utilizes Graph Attention Networks to analyze dynamic relationships between daily credit card transaction records, enhancing its ability to detect fraudulent activities. With the addition of differential privacy, the model allows multiple financial institutions to collaborate to refine the detection model without sharing sensitive data, thus improving adaptability and accuracy. The results are tested in two public datasets that show that FinGraphFL achieves accuracy rates of (Formula presented.) and (Formula presented.), significantly outperforming traditional methods. Building on these results, FinGraphFL sets the stage for future advances in real-time learning and global financial collaboration, ensuring simultaneous progress in security and privacy protections.
dc.language	English
dc.publisher	MDPI
dc.relation.ispartof	Mathematics
dc.relation.isbasedon	10.3390/math13091396
dc.rights	info:eu-repo/semantics/openAccess
dc.subject.classification	49 Mathematical sciences
dc.title	FinGraphFL: Financial Graph-Based Federated Learning for Enhanced Credit Card Fraud Detection
dc.type	Journal Article
utslib.citation.volume	13
pubs.organisational-group	University of Technology Sydney
pubs.organisational-group	University of Technology Sydney/Faculty of Engineering and Information Technology
pubs.organisational-group	University of Technology Sydney/Faculty of Engineering and Information Technology/School of Mechanical and Mechatronic Engineering
pubs.organisational-group	University of Technology Sydney/Faculty of Engineering and Information Technology/Engineering and IT Related HDR Students
utslib.copyright.status	open_access	*
dc.rights.license	This work is licensed under a Creative Commons Attribution 4.0 International License (CC BY 4.0). To view a copy of this license, visit https://creativecommons.org/licenses/by/4.0/
dc.date.updated	2025-10-09T04:33:33Z
pubs.issue	9
pubs.publication-status	Published
pubs.volume	13
utslib.citation.issue	9

Abstract:

In the field of credit card fraud detection, traditional methods often struggle due to their reliance on complex manual feature engineering or their inability to adapt to rapidly changing fraud patterns. This paper introduces an innovative approach called FinGraphFL, which merges graph-based learning with the principles of federated learning and improves security through differential privacy. FinGraphFL utilizes Graph Attention Networks to analyze dynamic relationships between daily credit card transaction records, enhancing its ability to detect fraudulent activities. With the addition of differential privacy, the model allows multiple financial institutions to collaborate to refine the detection model without sharing sensitive data, thus improving adaptability and accuracy. The results are tested in two public datasets that show that FinGraphFL achieves accuracy rates of (Formula presented.) and (Formula presented.), significantly outperforming traditional methods. Building on these results, FinGraphFL sets the stage for future advances in real-time learning and global financial collaboration, ensuring simultaneous progress in security and privacy protections.

Please use this identifier to cite or link to this item:

http://hdl.handle.net/10453/190331