Machine learning: The latest weapon in the fight against fraud
In recent months, machine learning has become somewhat of a buzzword, with major players such as Google highlighting the positive impact it can have for businesses.
Defined as a subset of artificial intelligence, machine learning focuses on the development of computer programmes that can teach themselves, and grow and change in response to new data. With 90% of the world’s data having been created in the last two years alone, the ability to develop automated processes to efficiently adapt to new information is invaluable.
For mobile operators, machine learning has the potential to drive huge benefits, in particular when it comes to tackling fraud. In 2016 alone, telcos are expected to face global losses of $294 billion, making it vital that they look to utilise all tools at their disposal to combat such a pressing issue, says Raul Gomes Azevedo, director Product Development, WeDo Technologies.
So how can machine learning fight fraud? For starters, fraud management involves identifying specific profiles and behaviour, and checking whether everything’s running as expected, or if there are any anomalies. Unlike other businesses, which are told to ignore anomalies, it is vital that fraud teams look to root these out.
Here, machine learning can play a key role, as algorithms can be developed and trained to monitor for such anomalies. For example, machine learning can identify unusual patterns and correlations from disparate data sources, going far beyond traditional rule-based fraud management. Advanced fraud management systems are even able to deliver a unique visualisation of verification results based on factors such as social network activities.
In addition, machine learning algorithms can enable the targeting of more complex risks, including both known and new, unknown threats, and with digitalisation continually breeding new and evolved fraud types, being able to identity and react to different, complicated threats as they arise is key to protecting revenues and reputations.
As a result of machine learning’s ability to efficiently and effectively identify and react to new threats, it can also help save valuable time for fraud management professionals. To fully fight fraud, human teams will always need to supervise any systems in place, and therefore rather than being something that will replace human skill, machine learning complements the efforts of fraud teams, freeing up their time to undertake other key tasks.
For example, machines can take over repetitive tasks, which are often disliked by fraud teams. An efficient fraud detection system should do the bulk of the heavy lifting, but escalate to human teams when additional insight is needed. Machine learning will also be particularly necessary as the Internet of Things continues to grow, fuelling exponential data growth.
With Gartner estimating that there could be over 20 billion connected things in use by 2020, the volume of data will be far too much for humans to process, let alone utilise the data to produce informed insights. In contrast, machine learning is ideally suited to fight fraud in a data rich age, as it has an inverse relationship with the size of datasets. This means that the larger the dataset, the more effective machine learning is.
Machines are also able to detect and recognise thousands of features and behaviours that could pose a risk to operators and identify them within microseconds, allowing operators to immediately react to potential threats. With such clear advantages, machine learning needs to become a key weapon in operators’ fight against fraud.
In particular, a new type of machine learning, deep learning, will be especially valuable. While machine learning’s main algorithm has been in existence for decades, new algorithms are now been developed each month, with deep learning algorithms being incredibly complex.
Deep learning software attempts to mimic the activity layers in the neocortex section of the brain, where thinking occurs. As a result, the software learns to recognise patterns in digital representations of sounds, images and other data. For fraud teams, this enables fraud management tools to be connected to services such voice recognition, biometrics etc, which will dramatically help to develop the rich profiles needed to tackle fraud, and mitigate new threats in a digital age.
Yet despite the clear advantages machine learning can bring to operators fighting fraud, it is still not widely used across the industry, with a recent survey from the TMForum revealing that only 23% of operators are currently using machine learning for fraud management.
However, this looks set to shift slightly, with 18% stating that they are considering complementing their current fraud detection approach with a data mining tool. When confronted with new threats and severe revenue losses due to fraud, the advantages machine learning can bring to operators are too important to ignore. They must therefore act now, and update their fraud management tools to maintain a competitive edge, and ultimately preserve their businesses.
The author of this blog is Raul Gomes Azevedo, director Product Development, WeDo Technologies
Comment on this article below or via Twitter: @ VanillaPlus OR @jcvplus
Digital services transformation – are we there yet? New analyst insight report
We’re delighted to bring you our specially-commissioned VanillaPlus Digital Transformation Insight. The Insight contains an exclusive Analyst Report written for VanillaPlus by Stratecast | Frost & Sullivan analysts Karl Whitelock and Troy Morley