Applying AI to secure the payments ecosystem

Description

Today’s cybercriminals are well organized and well resourced. Criminals invest in artificial intelligence and computing power the same way we do, helping to grow their market into a $600 billion enterprise-strength industry—or almost one percent of global GDP.

As the network connecting 3.3B cards, 54M merchants, and 15,600 financial institutions, Visa is uniquely positioned to detect data breaches affecting merchants in a timely manner. For merchants in the ecommerce space, where financials can be stolen via malware injected into the merchant websites, at scale, the threat posed by malicious actors continues to mount. Between 2015 and 2017, the percentage of merchant attacks occurring in the ecommerce channel went from 27 percent to 76 percent. And as two-thirds of breaches go undetected for a month or more, reducing the window of exposure is vital. Visa uses AI, big data, and its suite of risk products to monitor the ecommerce universe and detect breaches faster.

Chiranjeet and Shubham provide a high-level overview of the problem formulation, specialized data engineering, and several aspects of the model architecture, which includes traditional machine learning and deep learning, graph analytics, and semisupervised learning techniques such as generative adversarial network (GAN).