Credit card fraud in the modern world is becoming ever more sophisticated and it has become an increasingly difficult challenge for credit card companies to provide accurate yet timely fraud detection for their issuer banks.

With the advent of NoSQL and the plethora of choices for high speed, highly reliable data managment banks can now load and keep Terabytes of data in RAM and get access to this data at memory speeds. This approach has allowed two of the largest players in this space to reduce their Fraud Detection-to-Notification time from 45 minutes to under 5 seconds in one case and 1 second down to 60ms in the other.  Both can keep up with their peak processing load easily now. With more data available in memory, banks can not only run more complex fraud-detection rules faster, but can also enable predictive analysis with equally fast responses hence further reducing their Risk exposure and improving not just the performance but also the quality of their Fraud Detection.
In this session, Kunal Bhasin will introduce the audience to the in-memory computing techniques used by these two processing networksfor the detection of Fraud in real time.