Enhancing Cryptocurrency Fraud Detection: A Hybrid Model Combining Transformers and Graph Neural Networks

Sadda, Rama Subba Reddy (2025) Enhancing Cryptocurrency Fraud Detection: A Hybrid Model Combining Transformers and Graph Neural Networks. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (1MB) | Preview
Preview	PDF (Configuration Manual) Download (1MB) | Preview

Abstract

The rapid expansion of blockchain-based cryptocurrencies has fostered both innovation and a rise in financial crimes, including money laundering, ransomware payments, and illicit transfers. Traditional fraud detection approaches face challenges in this environment due to the decentralized, pseudonymous, and highly interconnected nature of blockchain transactions. This study leverages a hybrid model that integrates Transformer-based temporal feature extraction with Graph Neural Network (GNN) structural learning to improve the detection of illicit transactions. Using the Elliptic Bitcoin transaction dataset, classical classifiers (Logistic Regression, Naive Bayes, Random Forest) and deep learning baselines (Feedforward Neural Network, Graph Convolutional Network) are evaluated. While traditional and single-modality deep models capture either sequential or structural patterns, they fall short of leveraging both dimensions simultaneously. The hybrid architecture employs a Transformer-inspired autoencoder to learn high-dimensional temporal embeddings and anomaly signals, which are then processed by a GNN to model network context. Experimental results show the hybrid model achieving 94.9% accuracy, 98.1% precision, 98.1% recall, and a 97.3% F1-score—outperforming all baselines. These results highlight the effectiveness of combining sequence-level context with graph-structured reasoning, offering a robust and scalable framework for cryptocurrency fraud detection with potential for real-time and cross-platform applications.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Kelly, John UNSPECIFIED
Subjects:	Q Science > QA Mathematics > Computer software > Computer Security > Database security > Blockchains (Databases) T Technology > T Technology (General) > Information Technology > Computer software > Computer Security > Database security > Blockchains (Databases) Z Bibliography. Library Science. Information Resources > ZA Information resources > ZA4050 Electronic information resources > Databases > Distributed databases > Blockchains (Databases) Q Science > QA Mathematics > Computer software > Computer Security T Technology > T Technology (General) > Information Technology > Computer software > Computer Security H Social Sciences > HG Finance > Money > Digital currency
Divisions:	School of Computing > Master of Science in Data Analytics
Depositing User:	Ciara O'Brien
Date Deposited:	03 Jul 2026 09:51
Last Modified:	03 Jul 2026 09:51
URI:	https://norma.ncirl.ie/id/eprint/9455

Actions (login required)

View Item