Back to articles

Uses of machine learning & deep learning in fintech

7th September 2017
By Dan Lamyman
Co-Founder & Director
Business Intelligence & Advanced Analytics

The use of deep learning neural networks in FinTech has clear benefits across the financial spectrum. On the consumer side, we are looking at easier, safer payment methods and the facilitation of keeping track of our financial lives. On the industry side, however, there are some very big fish to fry, and these are some of the key areas in which deep learning is dominating.

Loan Applications

When it comes to lending, financial processing, underwriting, and decision making could be altered by deep learning technologies. The ability to process data in deeper, more sophisticated ways, and in a shorter amount of time, than humans are capable of means that deep learning algorithms can make more accurate decisions in a fraction of the time.

The major benefit is the ability for a deep learning algorithm to plough deep into data on an individual basis, testing for fraud, risk, and any important mitigating factors affecting loan approval. Such algorithms can analyse an applicant’s financial status in the context of current market trends and even relevant news items.

Security

FinTech companies like Aimbrain are tapping into the power of biometrics to step up authentication on financial service platforms. Aimbrain’s machine learning algorithm learns how a platform’s users interact with the platform online from all devices, on an individual level. The algorithm knows how much pressure you use to type, it knows your usual typing speed, what words you tend to type quicker, the way you navigate on a page. If the algorithm detects an anomaly from your usual behaviour once you’re logged in, you’ll be asked for facial or voice confirmation, as it will also do if you wish to make a large transaction.

Then, there are the machine learning systems used for cybersecurity. A company named DarkTrace uses AI that mimics the human immune system to defend enterprise networks from attack. It functions across all network types, from physical, virtualised, and cloud, through to IoT and industrial control systems.

Retrospective Reconciliation

Going back over previous statements and records to verify reliability is an important process for ensuring the accuracy of data. Nonetheless, when done manually, it’s a time-consuming, rather menial task. Machine learning algorithms can, of course, complete the job much faster, and with optimum accuracy, eliminating human error, saving time resources, and freeing up human labour for other tasks.

A perfect example is the Cube system developed by Duco. Duco Cube compares any type of data, in any format, in minutes. End users and subject matter experts can load data, compare it and find issues fast, without handovers or technology projects.

Trading

Simple AI programmes have been used in trading for years, which have been pre-wired to complete tasks set by traders. These include things like proposing a price for a certain share, making decisions based on a client’s risk profile, and so on. These aren’t terribly sophisticated systems, but deep learning is changing that.

The more data the deep learning system is fed, the more accurate its predictions will be. It is able to read data in ways that are pretty much invisible to the human eye. With this functionality, correlations and patterns that have never been observed before can be unearthed. Competition on Wall Street (etc.) therefore becomes down to who owns the most sophisticated deep learning system, rather than who can hedge the best bets on the market.

Regulation

The Treasury’s 2015 Budget Report identified RegTech (Regulatory Technology) as a whole separate sector of FinTech, demonstrating its intrinsic value to the banking industry.

Research by RegTech company, Suade, suggests that 75% of banks could be unprofitable as a result of the costs of regulatory implementation. These necessary regulations are designed to improve banks’ risk controls, maintain sufficient capital, and bring transparency to the financial sector. In order to meet these demands, however, banks need better technology.

Deep learning and artificial intelligence filters provide close to real-time insights, allowing banks to identify any regulatory problems in advance. Currently, it’s only really feasible to act on enforcement action after a breach has occurred. Pre-empting and preventing any breach of regulation before it occurs saves time, money, and resources.

To conclude…

FinTech is a many-headed hydra, harnessing the power of a range of different technologies to optimise Finance, both on consumer and industry level. Deep learning accounts for a significant proportion of the FinTech revolution, already demonstrating instrumental value across the board. Some may argue that entrusting such a key aspect of our society to machines could lead to problems in the future. However, all signs point to a positive outcome as the technology develops, with the evidence already plainly visible and growing fast.

Share Article

Get our latest articles and insight straight to your inbox

Hiring data professionals?

We engage exceptional humans for companies looking to unlock the potential of their data

Upload your CV
One of our consultants will contact you
to find out more about you