Mediaboss Marketing

PayPal head of emerging technology research, Hubert Le Van Gong.

PayPal

PayPal is looking to get in on the ground floor of a cutting-edge technology that could change the way the payments giant catches fraud and measures the creditworthiness of its customers.

Whether it's Goldman Sachs looking to speed up how it prices derivatives, or JPMorgan using quantum computing to test an algorithm that predicts options prices, top financial firms are exploring how and where the tech can be deployed.

Quantum computing, unlike traditional computing, uses a branch physics that runs on quantum bits rather than 1s and 0s. Because of this, quantum computing is helpful when executing large, complex calculations, like those in risk analytics or algorithmic trading.

The firm partnered with IBM in October 2020 to figure out how to use quantum computing to improve fraud detection, credit-risk operations, and overall security posture.

Early research shows quantum computing can be better than traditional computers in sweeping through large data sets and discovering patterns in data that can be indicative of fraudulent behavior or identifying credit-worthy individuals, Hubert Le Van Gong, PayPal's head of emerging technology research, told Insider.

Applying quantum computing to existing machine-learning capabilities could mean PayPal would improve its ability to detect fraud and save costs during the modelling process.

But it's a long-term play many of the benefits are theoretical and have yet to be proven.

"I wouldn't say this technology is going to detect fraud in a meaningful way anytime soon," Jay Gambetta, an IBM fellow and vice president of IBM Quantum, told Insider. "It's still very research-based," he added.

And even with "pretty aggressive" timelines in regards to quantum hardware and software, the technology won't be ready to implement until 2023, Gambetta added.

Even at such an early stage, it's a play the payments giant is ready to take on.

"It's not a matter of if, it's more a matter of when this is going to happen," Le Van Gong said. "The companies that are just sitting back and looking at it, waiting for it to become ready are going to miss out."

PayPal sifts through big, constantly changing data sets to detect fraudulent activity and make decisions around credit worthiness. However the data sets are large and can have millions of samples and up to 10,000 different properties like IP address, device type, or location, Le Van Gong said.

To cut down the number of properties and shave off computational costs of modelling, PayPal currently uses a method called "feature selection," Le Van Gong said. The process uses machine learning to pinpoint which properties are most useful in flagging fraudulent behaviors.

But even with feature selection, it's still an extremely complex, expensive, and time-intensive task to do with classic computers, he said.

"The scale at which PayPal operates in terms of machine learning is such that even classical computers, and the best computers you can find today, are going to be limited," Le Van Gong said. Quantum computers hold the promise of scaling beyond traditional computers when it comes to the number of data features and the size of the datasets, he added.

In addition to scale, quantum computers could help PayPal improve its prediction of important features and do so at a reduced cost compared with traditional computers, Le Van Gong said.

PayPal, which has been researching quantum computing for the past few years, is still in the learning stage of how the technology works and can integrate with classical computers.

The initiative is led by Le Van Gong's emerging technology research team, established in 2021, that explores the use of advanced technologies like cryptography and distributed-ledger security.

"It's still early in the process and it's very much humbling work," he added.

See the original post:
PayPal is experimenting with quantum computing to supercharge how it analyzes fraud and risk. An exec takes us inside the payment giant's playbook for...