Teradata (NYSE: TDC), a leader in data analytics , today announced that Danske Bank , a leading Nordic financial services company, has worked with Think Big Analytics to build and launch a cutting-edge AI fraud detection platform that is projected to deliver 100% ROI in its first year of production. The tool employs machine learning to analyze millions of banking transactions in real time and provide actionable insights into both fraudulent and credible activity . By significantly reducing the cost of investigation, Danske Bank is increasing its efficiency and savings.
“Fraud is a critical issue for banks, as there is evidence that criminals are becoming more clever and using sophisticated machine learning techniques, so advanced methods are needed to catch them,” says Nadeem Gulzar, Head of Advanced Analytics, Danske Bank. “The bank understands that this situation will only get worse in the future due to digitalisation and mobile applications. We recognise the need to use cutting-edge techniques to tackle this. Thanks to AI, we have already reduced false positives by up to 50% and have been able to reallocate up to half of the fraud detection unit to take on greater responsibilities.”
Danske Bank’s original fraud detection system was based on rules that were applied manually over the years. With a record of false positives reaching 99.5% of all transactions, the costs and time associated with investigation had become very significant. However, the fraud detection team was overworked and not being used effectively.
The Think Big Analytics team began working with Danske Bank in the fall of 2016 to tunisia phone number augment its advanced analytics team with expertise in using data to achieve greater business benefits. The joint team created a working environment with the bank’s existing infrastructure and then developed advanced machine learning models to detect fraud in millions of transactions annually. To ensure transparency and trust, the tool provides explanation and interpretation of blocked activity.
From a modelling perspective, fraud cases are still rare, one in a hundred thousand. The team has been able to reduce the false positives of the models by 50%. In addition, they are able to detect more frauds, currently around 60. Danske Bank's anti-fraud programme is the first to use machine learning techniques while developing deep learning models to test the techniques.
“Every bank needs an advanced and scalable analytics platform, as well as a digitalisation roadmap and strategy to introduce data science into the organisation,” says Mads Ingwar, Client Services Director at Think Big Analytics. “For online transactions, credit cards and mobile payments, banks need a real-time solution. The AI platform we have developed in collaboration with Danske Bank catches incoming transactions in less than 300 milliseconds. This means that when customers are in the supermarket shopping, the system can detect the transaction in real time and provide useful information immediately. We are starting to see this type of solution in organisations in the financial sector.”
Danske Bank will share details of its financial fraud countermeasures tool at the annual Teradata PARTNERS Conference in Anaheim, California, on October 23 at 11:30 a.m. Speakers include Ron Bodkin, Teradata's CTO of Services and Architecture, and Nadeem Gulzar, Danske Bank's Head of Advanced Analytics and Architecture.