Graph Neural Networks for Fraud Detection: Modeling Financial Transaction Networks at Scale

DOI

10.2991/978-94-6463-872-1_45How to use a DOI?

Keywords

Fraud Detection; Graph Neural Network (GNN); Anomaly Detection; Reinforcement Learning; Imbalanced Data; Financial Transaction; Yelp Dataset; Amazon Dataset; Multi-relational Graphs

Abstract

The worldwide economies are being seriously impacted by financial fraud, requiring proficient detection techniques able to spot changing and complex fraudulent activity. Conventional Machine Learning (ML) models and rule-based approaches among other traditional fraud detection systems find it difficult to scale, and adaptably capture relational fraud patterns in vast financial transaction networks, and we present a new Graph Neural Network (GNN)-based fraud detection model that improves both computational efficiency and detection accuracy in order to meet these issues. To develop expressive node representations, our method combines multi-hop neighborhood aggregation with attention methods, hence providing strong fraud detection. We also provide a hybrid detection system using community-based anomaly detection (Yelp-inspired) for detecting behavioral similarities and transaction-based embeddings (Amazon-inspired) to identify subconscious fraud movements. We use adaptive filtering systems and reinforce- ment learning-based neighbor selection to get above the constraints of highly imbalanced datasets, thus enhancing fraud detection performance and decreas- ing false positives. With 95.00% accuracy, 93.10% precision, 93.15% recall, and 93.20% AUC, experimental evaluations on real-world Yelp and Amazon datasets show that our model beats currently used GNN-based models including GCN, GAT, and GraphSAGE considerably. These findings confirm the success of our model in identifying large-scale fraudulent activity, providing a very attractive alternative for the prevention of financial fraud.

Copyright

© 2025 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Omkar Reddy Polu
AU  - Balaiah Chamarthi
AU  - Tanay Chowdhury
AU  - Azhar Ushmani
AU  - Pratik Kasralikar
AU  - Abdul Aleem Syed
AU  - Aashish Mishra
AU  - Sathish Krishna Anumula
AU  - Rethish Nair Rajendran
AU  - Manas Ranjan Mohanty
AU  - Nuzhat Noor Islam Prova
PY  - 2025
DA  - 2025/11/04
TI  - Graph Neural Networks for Fraud Detection: Modeling Financial Transaction Networks at Scale
BT  - Proceedings of the 2nd International Conference on Sustainable Business Practices and Innovative Models (ICSBPIM-2025)
PB  - Atlantis Press
SP  - 712
EP  - 729
SN  - 2352-5428
UR  - https://doi.org/10.2991/978-94-6463-872-1_45
DO  - 10.2991/978-94-6463-872-1_45
ID  - Polu2025
ER  -