Fighting money laundering with machine learning

A stack of credit cards

Money laundering is a serious financial crime that supposes a major threat to national public security. The aim of the project is improving the current anti-money laundering (AML) systems at the ING bank using machine learning technology. From the perspective of AI, the challenge is threefold. First, the extremely unbalanced data makes the traditional supervised machine learning paradigm unsuitable. Second, the sparsity of the financial networks (i.e., the fact that banks can only keep track of internal transactions) makes the AML task very hard since oftentimes criminal groups transfer money between different financial entities using intricate patterns. Third, regulators require banks to provide explanations for every reported client. That is, the AML model must not only be capable of assessing the probability that a client is conducting money laundering activities but must also provide explanations to support its decision. ING is very interested in the incorporation of machine learning technology to its systems to automatize the detection of money laundering. For this matter, they will provide access to their data as well as their analytics platform.


Ramon Rico Cuevas

Academic supervisors

Yannis Velegrakis, Ioana Karnstedt-Hulpus

Grant funding agency and (co-)funding non-academic partners