Reduce False Positives & False Negatives by Double Digits


On Premise & Cloud Solution for Realtime Adaptive Fraud Decision Model

We detect fraudulent card payments using our patent pending MicroSynapse algorithm, applying a dynamic Artificial Intelligence Machine Learning engine to generate, within 1-2 miliseconds on average, an accurate decision for each transaction.

RevenueStream manages to outperform leading industry solutions by significant margins, adapting itself to different customer bases –

Instead of manually setting rules as a response to events, the MicroSynapse engine will automatically calculate deep data such as  average spend per card / merchant / time slice or aggregate singular transactions into historically significant customer profiles.

This data is then used as features for the MicroSynapse engine, further enhancing performance.



Figure 1 – In this example we compare the results achieved with the openly available European Credit Card dataset on Kaggle, one run done with MicroSynapse disabled (top), another with it enabled (bottom).

Bear in mind that the data in both cases was cross validated on time series as opposed to randomly, to simulate real world conditions where unknown types of fraud are introduced to a previously trained model.