Relationship Between Machine Learning and Risk Detection

Machine learning is a branch of artificial intelligence, in which the computer uses algorithms to analyze data to identify patterns and create predictive models. Such models can be automatically adjusted with minimal human intervention so that they always deliver reliable and insightful results as soon as new data is available. This form of computer-based intelligence is changing the way international purchasing departments recognize and proactively address risks . This ranges from the security check of potential suppliers to routine monitoring of third parties in order to identify possible reputational, financial or strategic risks and to comply with the relevant laws.

Origin of machine learning

The term “machine learning” was coined in 1959 by Arthur Samuel, an American pioneer in the fields of computer games and artificial intelligence, who worked for IBM at the time. In the classic sense, machine learning applies the methods to statistical problems with clearly defined and structured data sets. This can be extended to deep learning, which emulates the complex learning processes in the neurons of the human brain. Algorithms are connected in series so that one algorithm filters the data according to a certain characteristic and the result is fed into the next algorithm, which in turn filters according to another characteristic and so on. However, these layers of features are not developed by human engineers,

A data revolution

Since machine learning requires large data sets and immense computing power, it is directly linked to the latest data revolution, explains Bart van Liebergen, Associate Policy Advisor at the Institute of International Finance, in his article Machine Learning: A revolution in risk management and compliance? “Certain elements of machine learning go back to the early 20th century,” says van Liebergen, “but the widespread use began when new computer technology and the easier availability of high-frequency data made it possible to represent complex, non-linear relationships in the model, and thereby significantly simplifying the use of machine learning. ”

Terms such as layered algorithms, structured data sets and predictive analytics may be difficult for laypersons to understand, but their value is still clear. Self-driving cars, personalized product recommendations on the Internet based on past shopping behavior, analysis of trends and data from wearables, i.e. wearable computer systems such as smartwatches with a monitoring function, are all applications that make our lives easier.

Uncover risks

Business applications accordingly increase the ability of companies to avoid risks; from forecasts of sensor failures in refineries and the rationalization of resource distribution in the oil and gas industry to tons of data on transactions and regulations or fraud prevention in financial services. Machine learning can also help identify risks related to bribery, corruption, or forced labor in supply chains.

“One area where machine learning has been used for more than a decade – and with considerable success – is the detection of credit card fraud,” said van Liebergen. “The historical transaction records show a variety of predetermined fraud properties that can be used to distinguish normal use of the card from fraudulent use.”

Machine learning provides valuable information because it enables complex patterns in data to be recognized more quickly than with human analysis. This enables companies to use both internal and external data to develop a broader understanding of risks among customers, suppliers and other third parties. With this information, companies can act proactively when threats to their reputation, compliance, sales or corporate strategy appear.

Why is third party data so important for machine learning?

Right from the start, companies have large amounts of data from their internal sources throughout the organization – from data from financial systems and CRM systems to data that capture intelligent machines in production processes. But the internal data alone is not enough to get the greatest possible amount of information from applications.

What types of third party data help to get a holistic understanding of the risks?

• Monitoring negative reporting in print media and online sources helps to identify possible threats in advance.
• Strategic risks and opportunities can be forecast with reports on countries and industries.
• Business data, including financial details, can reveal bankruptcy risks.
• The insight into company hierarchies helps to uncover the true beneficial owner.
• Sanctions, watchlists and lists of politically exposed persons (PEP) reduce the risk of bribery, corruption, money laundering or even the financing of terrorist organizations.
• Legal information helps to find out about previous legal disputes between potential customers or providers.

Machine learning is undoubtedly an important tool for advances in predictive analytics. But for corporate due diligence and risk management – including the numerous and ever-growing regulatory requirements for global operations – this tool works most efficiently when it receives the right kind of “fuel”.