Using Machine Learning to Fight Fraud and Create Insight - Driven Services

Husayn Kassai is CEO and co-founder of Onfido, whose machine learning (ML) technology helps companies verify identities online in seconds. In this article, we explain the use cases of ML in financial services and look at how Onfido’s ML technology can help financial institutions (FIs) to accelerate in a world of greater competition and growing regulatory pressure.

ML is the latest buzzword, it regularly appears alongside terms like Artificial Intelligence (AI), and advanced analytics. Many institutions claim to offer ML related services, but it’s hard to cut through the noise and understand what business benefits ML really delivers for FIs. So, what is ML? According to Arthur Samuel, ML gives “computers the ability to learn without being explicitly programmed.” It automates analytical model building through constructing algorithms that learn from and make predictions based on vast amounts of data.

ML - the use cases in Financial Services

FIs are developing multiple use cases for ML to help them address some of their biggest challenges including:

  • Customer engagement. ML can help uncover behavioural patterns that would suggest impending events. These events can relate to specific purchases, say, a house, or even changes of status, like an upcoming wedding or retirement. And it’s with these insights that FIs can offer services that add real value to the experiences of their customers.
  • Fraud prevention. ML is helping FIs recognise anomalies in customer behaviour so they can take preventative action and stop fraud or money-laundering activities in their tracks. Indeed, ML can uncover strange patterns and inconsistencies, pinpointing malicious intent much faster than ever before.

ML to verify identities in seconds

Onfido is helping FIs address governance and regulation around customer identities using ML. The company has developed its ML technology so that FIs can verify the identities of their customers online in seconds. Onfido takes advantage of widespread access to the internet, enabling identity validation using a webcam or smartphone camera. A person takes a photo of their identity document and then a photo or short recording of their face. Onfido’s ML technology then compares the image on the document with facial biometrics captured in the photo of the face while cross-referencing the identity document against international credit and watch list databases.

Husayn Kassai, CEO and co-founder of Onfido says, “Once a pattern is recognised and learnt by Onfido, it will scan all subsequent documents for this pattern. That means that one fraudulent document discovered may lead to many other documents being detected as fraudulent, if they follow the same fraudulent pattern.”

For Kassai, there are several benefits from using ML for identity verification. He says that it:

  • Increases the accuracy of the verification process.
  • Lowers the cost by reducing the need for manual checks.
  • Improves the customer experience with verification completed quickly.

As the Onfido system has verified millions of identities, the ML technology inside has enabled Onfido’s algorithm to improve - incrementally enhancing the accuracy and speed of its identity-checking service. “The performance of our ML technology is enhanced as we expand into new territories because it gives us access to new data sets,” says Kassai. “Today, we can verify 600 document types from 192 countries - helping us win international clients and scale into new areas.”

At the heart of ML power - the algorithm

For Kassai, Onfido’s big advantage is that it moved quickly to develop its algorithm. Therefore, the Onfido algorithm has had more time to operate, processing more data than competitors, and learning much more in the process.

He says there is a valuable lesson here for FIs that might be holding their ML investments back. The ability of ML to help a bank tackle regulation, increase operational efficiency and improve customer engagement depends on how much data its algorithm has been able to crunch. Hence, it’s those FIs whose ML technology is the most mature that will reap the greatest rewards.

Kassai also discounts the possibility of FIs that are slow to embrace ML eventually catching up. That’s because the front-runners will have perfected their algorithms and developed their internal skills to the point where not only are they tackling fraud and increasing operational efficiencies, but engaging with customers in such an effective way as to capture market share.

Accelerating ML development with Intel

FIs can quickly ramp up their ML capabilities if needs be. They can build out their ML platforms through the support of Intel, whose aim is to accelerate the potential of FIs in terms of ML, deep learning, and Artificial Intelligence (AI). Indeed, it’s through Intel’s wide portfolio of processors, memory, and storage and network solutions that financial services are being transformed worldwide.

FIs may find it beneficial to work with Intel to build out their broader ML capabilities - in areas such as customer engagement - and to engage the services of regtechs like Onfido for ML solutions to support more specialised areas, in this case, identity validation.

Whatever an FI’s ML strategy, time is of the essence. FIs need to evaluate right now how well they are using ML to reduce risk, increase operational efficiency and improve customer engagement. What’s clear is that ML is transforming the power of data to deliver actionable business intelligence. It is enabling FIs to tackle some of their biggest challenges, and will, ultimately, help them succeed in a world that is becoming increasingly competitive.

To find out more about Intel and how it is driving the adoption of ML across financial services visit click here, or to discover solutions about advanced analytics click here.

Democratising Algorithmic Trading Through a Cloud Strategy Based on Business Requirements

Quantiacs mission is to algorithmic trading accessible to the masses. In this article, the CEO explains Quantiacs’ cloud strategy and how it gives the company greater flexibility and efficiency, while at the same time helping the business meet its security needs.

Read more

Securing Data for Trusted Customer-centric Experiences

In this article, Heidi Rakels, CEO of GuardSquare discusses how FIs should be addressing their security requirements today and in the future.

Read more

Intel in Financial Services

Meet the demands of regulation and customers, minimizing risk while remaining profitable. Intel®-based solutions can help manage, structure and draw insights from the vast amounts of data and provide a scalable infrastructure to meet workload needs now and in the future.

Read more