A Hybrid Active Learning GraphBoost Approach for Anti-Money Laundering in Cryptocurrency Transactions

-, Chaolu (2025) A Hybrid Active Learning GraphBoost Approach for Anti-Money Laundering in Cryptocurrency Transactions. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (2MB) | Preview
Preview	PDF (Configuration Manual) Download (804kB) | Preview

Abstract

Cryptocurrencies present significant anti-money laundering challenges due to their pseudonymity, evolving laundering tactics, and severe class imbalance. Traditional rule-based and static machine learning methods struggle to adapt to these dynamic transaction networks. This research introduces a Hybrid Active Learning GraphBoost Framework to address these limitations. The approach integrates graph neural networks with active learning to model complex transaction topologies while minimizing labeling costs. Key innovations include: Active learning strategies to prioritize high-value transactions for expert annotation, reducing labeling effort while improving detection. GraphBoost Model, a stacking ensemble combining Graph Neural Networks with Random Forest via a Logistic Regression meta-learner, leveraging both graph structural and node feature information. Comprehensive benchmarking showing GraphBoost outperforms standalone models, achieving a 0.9885 F1-score, 0.9957 AUC-ROC, and 0.9794 AUC-PR on the Elliptic Bitcoin dataset. Experiments confirm the superiority of graph-aware models over feature-based baselines. The Active Learning framework enhances a base GCN model’s F1-score from 0.9478 to 0.9492 with only a 2.68% expansion in labeled data. The refined GraphBoost Model identifies 11, 169 potentially illicit unlabeled transactions, demonstrating scalability and robustness for real-world AML deployment.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Cosgrave, Noel UNSPECIFIED
Subjects:	H Social Sciences > HG Finance > Money > Digital currency H Social Sciences > HG Finance > Fintech T Technology > T Technology (General) > Information Technology > Fintech Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Machine learning
Divisions:	School of Computing > Master of Science in FinTech
Depositing User:	Ciara O'Brien
Date Deposited:	24 Jun 2026 10:11
Last Modified:	24 Jun 2026 10:11
URI:	https://norma.ncirl.ie/id/eprint/9388

Actions (login required)

View Item