The patents are for its Fragmentation Engine and Sandbox Layered State, and Featurespace leverages them to combat financial crimes associated with card-based transactions, account-to-account (A2A) transactions and fraudulent apps that are falsely used to check customers’ details when opening either a bank or customer account.
Due to the expected global recession, financial crime is believed to increase, thus making the need for patents timely. The announcement details that UK Finance predicts ‘an epidemic of fraud’ throughout 2023, whereas the United Nations estimates that money laundering costs as high as USD 2 trillion on a yearly basis, thus undermining economic growth and financing organised crime.
With its patented technology, Featurespace aims to reduce instances of fraud and financial crime and prevent transaction fraud in real time.
Decisions must be quick so that they can interdict real-time payments;
The technology must keep an extensive number of behavioural profiles updated with the latest transaction information;
And the technology must be robust as it is leveraged within the global payments flow.
Because of this, reliable high throughput and low latency event processing are required. Featurespace’s fragmentation engine patent aims to address these concomitant event processing requirements and works by having local copies, or caches, of customer behavioural profile data kept on each processor node and a global all-reduce process performed frequently to keep each processor’s view in sync. The solution is believed to help benefit the payments industry by improving the quality of data leveraged by the company’s ML models and resulting in risk detection accuracy at scale.
A/B testing of change versions;
And backfill and promotion of new versions of behavioural profiles.
Based on announcement details, financial institutions use this to roll out improvements to their fraud prevention analytics progressively, such as by adding device information and profiles to an existing set of profiles based solely on transaction data. Furthermore, they can also use it to run two versions of behavioural analytics concomitantly, being enabled to run a live A/B test of the analytics and promote the more effective version to production.
The invention achieves operations with hardware efficiency, which is believed to make them suitable for production systems that run on minimal hardware on-premises. To achieve this, the solution leverages memory management and search processes that involve copy-on-write snapshot caches and time-aware overlays.
Featurespace’s most recent patents are intended for use by regulated financial services, payment service providers, banks and businesses in the international fight against financial crime and fraud, with the patent technology being designed with legal and regulatory frameworks in mind of the likes of the EU’s Artificial Intelligence Act and the Bank of England’s Artificial Intelligence and Machine Learning regulation.
Every day we send out a free e-mail with the most important headlines of the last 24 hours.
Subscribe now