CubeFlow: Money Laundering Detection with Coupled Tensors

Abstract

Money laundering (ML) is the behavior to conceal the source of money achieved by illegitimate activities, and always be a fast process involving frequent and chained transactions. How can we detect ML and fraudulent activity in large scale attributed transaction data (i.e. tensors)? Most existing methods detect dense blocks in a graph or a tensor, which do not consider the fact that money are frequently transferred through middle accounts. CubeFlow proposed in this paper is a scalable, flow-based approach to spot fraud from a mass of transactions by modeling them as two coupled tensors and applying a novel multiattribute metric which can reveal the transfer chains accurately. Extensive experiments show CubeFlow outperforms state-of-the-art baselines in ML behavior detection in both synthetic and real data. *Xiaobing Sun, Jiabao Zhang, and Qiming Zhao contribute equally as the first authors. Shenghua Liu is the corresponding author.The work was done when Xiaobing Sun and Qiming Zhao were visiting students at ICT CAS, who are separately from

Anomaly Detection in Graphs of Bank Transactions for Anti Money Laundering Applications

“…This pattern corresponds to what in the AML community is known as the placement-layering-integration model [8], with each activity being carried out by the source, middle and destination nodes respectively. Intermediate accounts are also considered in [9], which uses a tensor approach to modelling tripartite patterns relevant to money laudering. • A cycle is a classic fraud pattern, especially if the amount transferred over it is nearly constant; in [10], cycles with various lengths are detected in real time, using a hot point index to speed up the search.…”