Research Article
Community Detection in Financial Networks for AML Using GNNs
@INPROCEEDINGS{10.4108/eai.28-4-2025.2358093, author={M.S. Minu and S. Eswaran and Janisa Ria D and V.B. Shreesha and T. Avinash}, title={Community Detection in Financial Networks for AML Using GNNs}, proceedings={Proceedings of the 4th International Conference on Information Technology, Civil Innovation, Science, and Management, ICITSM 2025, 28-29 April 2025, Tiruchengode, Tamil Nadu, India, Part II}, publisher={EAI}, proceedings_a={ICITSM PART II}, year={2025}, month={10}, keywords={graph neural network (gnn) graph convolutional network (gcn) graph attention network (gat) extreme gradient boosting (xgboost) anti-money laundering (aml)}, doi={10.4108/eai.28-4-2025.2358093} }- M.S. Minu
S. Eswaran
Janisa Ria D
V.B. Shreesha
T. Avinash
Year: 2025
Community Detection in Financial Networks for AML Using GNNs
ICITSM PART II
EAI
DOI: 10.4108/eai.28-4-2025.2358093
Abstract
Illegal money storing such as financing terrorism, corruption, and organized crime all pose a profound threat to our world economy by laundering money. Anti Money Laundering Protocols fail to keep up with sophisticated methods of scam detection. Using IBM's illicit patterns dataset from Kaggle, this paper investigates a novel approach to AML detection by utilizing machine learning Designed for high-volume transactions in digital economy. The work presented aims to develop and implement machine learning model to prioritize financial transactions for manual investigation, leveraging the IBM AML dataset. Using semi-supervised learning, the model analyzes transaction patterns, sender-receiver profiles, and historical behavior to improve detection accuracy. Additionally, the project explores regulatory frameworks and compliance strategies adopted by key financial agencies to enhance AML effectiveness. The findings aim to provide actionable insights to strengthen existing AML systems and support regulatory compliance. Building on existing literature on automatic detection of financial crimes, this paper also provides recommendations for more efficient and flexible systems for AML work.
