Fighting Fraud with Machine Learning

Data breach. Identity theft. Account takeover. These are just a few of the risks of living in a digital world, where fraud has been on the rise—a trend that’s hardly a surprise in light of the growth of e-commerce, not to mention the move toward a cashless society.  

Skip to main content
Main Content

Fighting Fraud with Machine Learning

May 21, 2018

Fighting Fraud with Machine Learning

Data breach. Identity theft. Account takeover. These are just a few of the risks of living in a digital world, where fraud has been on the rise—a trend that’s hardly a surprise in light of the growth of e-commerce, not to mention the move toward a cashless society.

Of course, the payments industry is not immune. In fact, the industry has seen a rise in card-not-present (CNP) fraud since the introduction of chip-enabled cards, which have made it much more difficult to perpetrate fraud at the point of sale. As a result, fraudsters have moved online, where they have found it easier to execute their schemes.

In fact, according to the Global Fraud Attack Index, which analyzes fraud attempts on e-commerce merchant web sites, total fraud rose 5.5 percent in Q2 2017, as compared to Q2 2016. Not surprisingly, high-dollar transactions were more likely to be fraudulent. In fact, the fraud rate for transactions over $500 was measured at 11.64 percent, 22 times higher than the fraud rate for transactions of less than $100.

At the same time, though, better tools are now emerging to help fight fraud, many of which utilize artificial intelligence (AI)--and more specifically, machine learning.

What is machine learning?

Dictionary.com defines machine learning as “a branch of artificial intelligence in which a computer generates rules ... based on raw data that has been fed into it.”

Machine learning has been around for decades. In fact, in 1959, artificial intelligence pioneer Arthur Samuel defined it as a “field of study that gives computers the ability to learn without being explicitly programmed.”

Yet there has been an explosion in the use of machine learning in recent years. For example, has an e-commerce site offered you product suggestions while you’re shopping online? Those suggestions probably take into account your account profile and previous purchases—and represent the use of machine learning.

Machine learning has also enabled self-driving cars, a technology that wouldn’t work if a self-driving car could not learn the rules of engagement between the vehicle and the road.

Factors enabling the rise of machine learning

One question often asked is why machine learning has suddenly become a bigger part of our world in the twenty-first century when it has been around for a long time? According to McKinsey & Company Financial Services, there are two primary reasons. First, “companies, institutions, and governments now capture vast amounts of data as consumer interactions and transactions increasingly go digital. At the same time, high-performance computing is becoming more affordable and widely accessible.”

As noted above, machine learning has more than its share of applications for the data-rich payments industry. But one way in which it is playing an increasingly important role is in fraud prevention, an obvious and important use of machine learning.

The goal of fraud prevention, in this context, is to help businesses to reduce fraud risk while still allowing legitimate customers to do business. That requires putting a stop to fraudulent transactions in real-time without blocking legitimate transactions (sometimes referred to as “false positives”).

In the past, systems used broad, human-generated rules in an attempt to limit the number of fraudulent transactions, which tended to result in too many false positives. Machine learning does a much better job. It “uses artificially intelligent computer systems to automatically learn, predict, act and explain without being explicitly programmed. Simply put, machine learning eliminates the use of preprogrammed rule sets,” explains Feedzai’s Primer to Machine Learning for Fraud Management, which is very useful, as it allows for the leveraging of fraud scoring. And banks, card issuers and merchants that can “outsmart risk” gain a significant competitive advantage.

How machine learning helps combat fraud

Naturally, machines that learn can be especially useful, because with all of their processing power, they’re able to quickly highlight or find patterns in data that humans would have otherwise missed. At the risk of oversimplifying, the process generally involves:

  • the system creating profiles based on gathered data
  • then algorithms evaluate transactions for fraud risk
  • the system takes into account historical data and streaming data to make a ‘decision’ in real-time 

A machine learning behavioral analytics approach builds a pattern of normal behavior for customers and then spots not only when that behavior changes, but automatically updates itself, and notes the learning as a new type of behavior. This change in behavior can alert a fraud team to review the new behavior for possible fraud attack.

How should a machine learning system operate?

The elements needed for a machine learning system calls on businesses to identify the business problem they are looking to solve, operationalize the solution, develop the technical components of the solution and then improve the system over time.

To be sure, fintech companies have been making tremendous inroads using machine learning in recent years. Consider Decision Intelligence™ by Mastercard which uses AI to solve “a major consumer pain point of being falsely declined when trying to make a purchase,” said Ajay Bhalla, Mastercard’s president of enterprise risk and security, in a press release announcing the use of the technology across the brand’s global network.

When looking to pinpoint cyberattacks, the latest developments in machine learning may be the best defense in the future against combatting card-not-present fraud.

Contact Us
About Our
Merchant Services

Get your Free Quote, Now!

After you have submitted your information, a TSYS representative will contact you.

All fields are required to submit form. Your information is private and secure. We do not accept adult businesses

Customer Support Form