HNC Software and MasterCard to Use Neural Network Technology for Fraud Detection

HNC Software Inc. and MasterCard International say they plan to jointly develop a system to help MasterCard members better detect fraudulent credit card transactions. The system will use HNC's neural network modeling technology, combining cardholder, merchant and geographical data to give MasterCard members a comprehensive fraud prediction model it says is unique in the industry.

The first module of the new risk-predictive service should be available in the fourth quarter of this year for MasterCard and its member institutions. The product can be integrated with an issuer's existing system or used as a stand-alone fraud detection service.

"Although both MasterCard and HNC Software have played an important part in the decline of credit card fraud over the last three years, there is still room for further inroads against this crime," says Michael Thiemann, president, HNC Software Financial Solutions."

According to Christopher Thom, executive vice president, MasterCard International, "This relationship will enable our companies to bring predictive model products to market rapidly, leveraging HNC's model development expertise and MasterCard's global product distribution capabilities to maximum advantage. The fact that MasterCard's global authorization system will be used to provide real-time notification is a major differentiating factor."

The MasterCard fraud-predictive system will constantly update information with each new transaction, building a detailed profile of each merchant and cardholder. This exclusive use of dynamic merchant profiles to supplement the transaction-based account profiles is designed to improve the predictive precision of the model. By using a variety of models of behavior for each global region, the system aims to ultimately allow members to examine similar patterns in a given set of countries.

Pilot Program Underway
A dozen MasterCard members are participating in a pilot program for the credit card fraud detection module. Firstar, based in Wisconsin, used the product to catch previously undetected fraudulent activity that could have resulted in an average loss of $5,100 per account identified.

"The complementary nature of the MasterCard product and our in-house system allowed us to confirm more cases of fraud with very little duplication - only one in five cases were identified by both systems," says Pat Wesner, senior vice president, Firstar. "The wealth of merchant data the MasterCard system utilizes is an advantage in our anti-fraud efforts."

"We view this as just the beginning of a relationship with a world of potential," says Thiemann. "We look forward to working with MasterCard to continue to offer its membership the most comprehensive fraud technology and information available."