Faster payments and certainly real time payments are new. The payments industry has hardly begun to understand its uses, opportunities and impact to other payment types. The industry has also only begun to understand the implications of fraud in these faster channels. In an article and interview on Bank Info Security, experts highlight the ways fraud mitigation is being handled in environments where transactions are executed and gone in a matter of seconds. Not surprisingly it requires a lot of data and machine learning to turn the data to action:
As banks in the U.S. and Australia grapple with how to effectively launch faster payments, more will turn to big data and machine learning to help better manage expected upticks in fraud, says John O’Neill Jr., director of financial crime and analytics at DarkTower, formerly Queen Associates.
The important thing for businesses to remember, especially banks, is that when it comes to machine learning, it actually has to learn,” he says. “So if I can equate that to a child: A child is given a bike and then they’re able to ride it. But they first have to learn how to ride it. So in this big data realm, where we’re trying to capture a lot of information and actually put it in so that it can be used to find trends, you have to teach it how to actually find fraud.”
And not all fraud is perpetrated equally, he adds, which means more “teaching” and “learning” has to take place. “There are many different types of fraud that can be committed – everything from check fraud to online payment fraud to wire transactions. Any of these [big data/machine learning] implementations really need to understand what those things are that make up fraud and how fraud is committed in these specific environments.”
Overview by Sarah Grotta, Director, Debit and Alternative Products Advisory Service at Mercator Advisory Group
Read the quoted story here