This article first appeared in Data Science Briefings, the DataMiningApps newsletter. Subscribe now for free if you want to be the first to receive our feature articles, or follow us @DataMiningApps. Do you also wish to contribute to Data Science Briefings? Shoot us an e-mail over at briefings@dataminingapps.com and let’s get in touch!
Contributed by Bruno Deprez
Money Laundering
The end goal of any criminal activity is to generate money. To be able to spend their money, criminals resort to money laundering, with the sole intend of hiding its illegal source. Some estimates put the amount of laundered money annually at USD 2 trillion, amounting to 5% of global GDP [8]. Since money will pass through the financial system at some point, regulation is put on banks and other financial systems. A first line consists of ‘know your customer’ (KYC) practices, screening clients when they want to do business with the bank. A second line deals with continuous transaction monitoring, where existing clients are being analysed to see if any suspicious patterns start to arise. Both KYC and continuous monitoring require the institutions to have a thorough understanding of money laundering.
Money laundering in general happens in three main steps [5]. During placement, dirty money enters the financial system, often in places where banking regulations are less strict. During layering, many different transactions are done to obfuscate the real origin of the money. This can also involve altered financial statements exaggerating one’s business revenue. During the final step, integration, the money enters wider circulation. The full scheme involves many different transactions and actors. Hence, capturing these interactions using network analytics provides a real benefit compared to just more classic features.
Network Analytics
Network analytics is a broad term that covers different methods to analyse network data to extract meaningful features or other knowledge. Initially, it dealt with summarising the relative importance of individual actors in the network. A very famous example is PageRank [7] behind the Google search engine where each website is seen as a node connected to others through hyperlinks forming the network.
More recently, deep learning methods have been developed to learn directly on the network itself, giving state-of-the-art performance [1, 3, 2]. These methods have recently been used by Google Deepmind to simulate novel crystal structures and their properties, pushing the boundaries of material science [6].
Open Problem for Money Laundering
Since anti-money laundering lends itself nicely to network analytics, many different approaches have already been applied. However, these are often based on general metrics, not tuned to the specific money laundering problem. Applications of machine and deep learning are not mainstream yet. The main hurdle is the lack of explainability [4]. In the end, when someone is suspected of money laundering, an analyst comes in and files a suspicious activity report (SAR) to the regulator. Therefore, there is a real need of good understanding of why a person’s finances are suspicious.
Conclusion
Money laundering is a pervasive global problem of criminals trying to hide the origin of their illegal funds to expand their activities. Given its specific nature, it lends itself well to the application of network analytics. However, given that interpretability is very important, current research often focuses on extending existing models with well-known network centrality measures, without necessarily tailoring these metrics to anti-money laundering.
References
- [1] Hongyun Cai, Vincent W Zheng, and Kevin Chen-Chuan Chang. A comprehensive survey of graph embedding: Problems, techniques, and applications. IEEE transactions on knowledge and data engineering, 30(9):1616–1637, 2018. doi: 10.1109/TKDE.2018.2807452.
- [2] Palash Goyal and Emilio Ferrara. Graph embedding techniques, applications, and performance: A survey. Knowledge-Based Systems, 151:78–94, 2018. ISSN 0950-7051. doi: https://doi.org/10.1016/j.knosys.2018. 03.022. URL https://www.sciencedirect.com/science/article/pii/S0950705118301540.
- [3] Will Hamilton, Zhitao Ying, and Jure Leskovec. Inductive representation learning on large graphs. 30, 2017. URL https://proceedings.neurips.cc/paper_files/paper/2017/file/5dd9db5e033da9c6fb5ba83c7a7ebea9-Paper.pdf.
- [4] Dattatray Vishnu Kute, Biswajeet Pradhan, Nagesh Shukla, and Abdullah Alamri. Deep learning and explainable artificial intelligence techniques applied for detecting money laundering–a critical review. IEEE Access, 9:82300–82317, 2021. doi: 10.1109/ACCESS.2021.3086230.2
- [5] Michael Levi and Peter Reuter. Money laundering. Crime and justice, 34 (1):289–375, 2006. doi: 10.1086/501508.
- [6] Amil Merchant, Simon Batzner, Samuel S. Schoenholz, Muratahan Aykol, Gowoon Cheon, and Ekin Dogus Cubuk. Scaling deep learning for materials discovery. Nature, 624(7990):80–85, 2023. doi: 10.1038/s41586-023-06735-9. URL https://doi.org/10.1038/s41586-023-06735-9.
- [7] Lawrence Page, Sergey Brin, Rajeev Motwani, and Terry Winograd. The pagerank citation ranking: Bringing order to the web. Technical report, Stanford InfoLab, 1999.
- [8] United Nations Office on Drugs and Crime. Money laundering. https://www.unodc.org/unodc/en/money-laundering/overview.html
