Only 16 percent of banks polled in a recent global IBM Institute of Business Value study could detect fraud as it is attempted. Once new schemes were identified and confirmed, it is estimated that countermeasures typically require over four weeks to deploy1. Existing fraud detection techniques are constrained by the dependency on specialized “black box” models that are hard to understand, explain, and adapt. As the payments industry continues to evolve with faster, alternative and mobile payments innovations and the expanding use of chip & PIN, fraudsters have become more technologically sophisticated and organized, adapting their fraud techniques more quickly.
IRIS provides a real-time fraud analytics engine that leverages machine learning to generate rapid anti-fraud models while also supporting the creation and modification of ad-hoc models, proven successful on various sized payment platforms. Acting as a “virtual analyst” to help identify new fraud patterns, IRIS applies machine learning models to help analysts detect fraud and act quickly to help reduce fraudulent events. IRIS serves to bridge the gap between expert-driven rules and traditional predictive modeling by applying artificial intelligence and cognitive techniques to partner with human experts in suggesting best fit analytics interactively, while testing and deploying models with real production data, as it happens and without downtime.
“The cognitive computing approach unleashes a new paradigm in fighting fraud,” said Alistair Rennie, general manager, Industry Solutions, IBM. “By integrating IRIS Analytics with IBM’s counter fraud technology, we will help organizations more accurately detect fraud at scale and speed so that they are in a position to implement countermeasures quickly, with more control and transparency, while at the same time assists with dramatically lowering false positives.”
IRIS is in use by leading banks and payment processors throughout the world. For example, the French payment card switch: e-rsb, opérated by STET, utilizes IRIS for 5.5 billion annual credit and debit card transactions. “With an average response time of less than five milliseconds per transaction even during peak periods when we are processing over 750 transactions per second, IRIS enables us to detect potential fraud without adding any notable overhead to our service,” states Pierre Juhen, Deputy CEO of STET. “In addition, we are able to respond to newly identified fraud patterns by deploying new countermeasures in a few hours without taking down the system.”
“Defenses against financial crime are in critical need of innovation and improvement. As the payments industry evolves with new payments methods such as chip and PIN, mobile payments and immediate payments, the ability for financial institutions to accurately make decisions about what is suspicious and what is legitimate before the payment is executed is required,” said Constantin von Altrock, Chief Executive Officer of IRIS Analytics. “The combination of IRIS technology with IBM’s Counter Fraud capabilities creates a comprehensive solution for real time payment fraud prevention.”
—
Image: Fraud by © Volkova Vera via Shutterstock
