Applications of machine learning in the identification, measurement and mitigation of money laundering

Abstract

The cost of financial crimes compliance continues to grow, locked in step with increasing regulatory expectations and volumes of low-productivity work items. Financial institutions cannot afford to wait for entirely new paradigms and instead are investing in solutions that provide near-term relief and can orient institutions towards the future. Technologies like artificial intelligence and machine learning (ML) — well entrenched in applications like credit risk modelling and fraud detection — are gaining traction within the broader financial crimes domain, and anti-money laundering (AML) in particular. To obtain the business value of these ML and other technologies, financial institution managers need the toolset to succinctly understand these methods and assess what approaches are appropriate and effective for their institutions. The twofold goals of this paper to equip institutional stakeholders with this information are: 1. Describe the high-level applicability of ML to AML, with a focus on transaction monitoring. 2. Provide an overview of the AML ML practices that are already in place within the industry; are on the immediate horizon; or are promising opportunities actively being investigated for the future.