Using Machine Learning to Detect Money Laundering
In this talk, Maria (Mahtab) Kamali, Data Scientist at Thomson Reuters, will present new machine learning methods employed to discover money laundering patterns.
Money laundering is the process of transferring profit from crime and illegal activities into legitimate assets. Based on the United Nations Office on Drugs and Crime, 2 to 5% of global GDP, or $800 billion - $2 trillion, is laundered globally on an annual basis. The laundered money often finances drug trafficking, human trafficking and terrorist activities.
Advanced analytic techniques are increasingly being employed to identify and reduce illegal activities such as money laundering. Machine Learning (ML) is playing an increasingly important role by way of two main mechanisms: transaction behavioural pattern analysis and network structure. Many financial institutions combine these two mechanisms to construct a rule-based system to flag suspicious transactions. A challenge is that these systems can generate significant false positives which require tedious resource-intensive investigations. Such rule-based systems are also challenged when seeking to detect new patterns and/or activities. Modern data mining and machine learning methods can help financial institutions reduce system-generated false positives.