Beginning Anomaly Detection Using

 Finance  and Insurance

In the finance and insurance industries, anomaly detection can be used to detect 

fraudulent claims, fraudulent transactions such as transfer of money in and out of the 

country, fraudulent travel expenses, and the risk associated with the specific policy or 

individual, etc. The finance and insurance industries depend on the ability to target 

the right consumers and take the right amount of risk when dealing with finance and 

Figure 8-11.  Fake news on Facebook

Chapter 8   praCtiCal Use Cases of anomaly DeteCtion

---

309

insurance. For instance, if they already know that a specific area is prone to forest fires 

or earthquakes or very frequent flooding, the insurance company insuring your home 

needs to have all the tools that they can get their hands on to quantify the amount of risk 

involved when writing the policy for homeowner insurance.

Anomaly detection can also be used to detect wire fraud where a large amount 

of money is transferred in and out of the country using several different accounts, 

something extremely difficult for human eyes to manually glance over and figure out 

considering the massive volume of transactions that can take place every hour. This is 

feasible because AI techniques can be trained on very large amounts of data to detect 

very new and innovative wire fraud beyond the capabilities of any human or many of the 

statistical techniques that have been in place for decades. Deep learning does solve a very 

big problem in the financial and insurance industries, and with the advent of graphical 

processing units (GPUs), this is becoming a reality in many of the very hard- to- crack use 

cases. Anomaly detection and deep learning can be used together in order to serve the 

needs of the business. Figure 

8-12

 shows the mortgage loan fraud reporting trend.