Mastercard Enlisting AI, ML in Fraud Fight

Mastercard – the payment solutions giant that operates across markets of all sizes for government, business, and consumers – has a highly vested interest in fighting financial fraud which poses dollars and cents risks to the company, its customers, and the quality of their experience with the firm.

With a bottom-line mission of stopping fraud in its tracks without causing delays in processing transactions or diminishing user experience, Mastercard is harnessing advanced technologies including artificial intelligence (AI) and machine learning (ML) solutions closer to the edge to help government and industry organizations get their business done faster and more securely.

Fighting transactional fraud is a challenge for any company, but with the scale at which Mastercard works – billions of transactions per day totaling trillions of dollars per year – the problem is even more challenging. But when everything is running at speed, similarly to how Mastercard operates, AI makes a significant impact on the bottom line.

A look at some of the numbers involved shows why further use of AI tech has found a home at the payment firm. The company needs to process 165 million transactions per hour, apply 1.9 million rules per transaction to prevent fraud, and do all this in milliseconds to prevent fraud in real-time.

“Our goal is to stop fraud in its tracks without disrupting or delaying legitimate transactions. We want to move intelligence closer to the edge to get closer to our customers,” said Nick Curcuru, vice president of Big Data Practice at Mastercard, in a case study.

To identify and prevent fraudulent transactions, Mastercard leverages AI and machine learning (ML) technologies that run on HPC systems to analyze large data sets at lightning-fast speeds on a secure, payment card industry–certified system from Dell Technologies.

According to Curcuru, Mastercard applies the 1.9 million fraud detection rules – nearly instantaneously – to the 165 million transactions per hour utilizing AI and ML algorithms running on Dell Validated Designs.

Dell Validated Designs for AI/ML solutions shorten deployment timelines with solutions designed using Dell and NVIDIA-proven AI expertise to deploy an AI solution optimized to produce ever-faster insights.

“This fraud detection system uses supervised learning to look for established fraud patterns and unsupervised learning to identify emerging fraud patterns in real-time,” he said.

With every transaction, the algorithms examine a customer’s buying habits, geographic location, travel patterns, and real-time data on account usage. Each transaction is analyzed to determine what is a valid or fraudulent transaction.

The result of these efforts is a more trustworthy transaction experience, plus real-time barriers “to stop criminals who try to exploit vulnerabilities in payment systems,” Curcuru said.