CA Technologies Debuts Risk Analytics Network to Reduce Online Fraud

The neural network-based analytics model will make use data inputs from multiple credit card issuing banks to identify potentially fraudulent online transactions.

online fraud

CA Technologies announced its Risk Analytics Network service on May 4, providing new capabilities to help reduce online fraud. The CA Risk Analytics Network makes use of machine learning and neural network model techniques to rapidly identify potentially risky devices and transactions.

"With EMV chip cards what we have started to see is that fraud is shifting from being within stores to going online," Terrence Clark, general manager for CA Technologies Payment Security solutions, told eWEEK.

CA already had a payment fraud offering in the market called CA Technology Payment Security Suite. What the Payment Security Suite does is look at various attributes of a cardholder's behavior, including where the user device is located, to identify potential fraud, he said. The new CA Risk Analytics Network complements the Payment Security Suite with new insights and context about potential fraud.

"Our data scientists realized that when fraud attacks occur, it's likely that multiple fraudulent transactions will be attempted from the same device," Clark said. "Fraudsters will also use multiple credit cards from different issuing banks to carry out attacks."

CA's analysis also revealed that attacks tend to occur in rapid, short bursts of time, such that within a few minutes, hundreds of transactions can be attempted. The CA Risk Analytics Network is a service that can make use of device information across different credit card issuers.

"So now we can look at card history, the device behavior and the device history simultaneously," Clark said. "We have expanded the fraud scope investigation across multiple issuers, instead of just focusing on a single credit card issuer."

Clark added that the broader scope of the Risk Analytics Network provides an expanded pool of knowledge about fraud. The fraud model is updated in real time, providing rapid assessment of emerging threats.

"The shift is that instead of just looking at the card history, we're looking at the device history and making use of data across our participating banks to help one another," Clark said.

The way the system works is CA takes the information about a transaction and feeds it into a self-learning module that creates thousands of attributes that are analyzed and assessed in real time by the Risk Analytics Network.  

"The system is self-learning so every new transaction helps to improve the model over time," Clark said. 

The neural network model that powers the CA Risk Analytics Network was built by CA and isn't based on any open-source project, according to Clark. While the data science behind the technology is currently being used to help mitigate fraud, he said future applications could extend the technology to help other areas of security, including anomaly detection.

Sean Michael Kerner

Sean Michael Kerner

Sean Michael Kerner is an Internet consultant, strategist, and contributor to several leading IT business web sites.