Unlike rules-based systems, which are fairly easy for fraudsters to test and circumvent, machine learning adapts to changing behaviors in a population through automated model building. With every iteration, the algorithms get smarter and more accurately find activities that represent risk to the firm.
