Fraud is a significant and evolving challenge for the financial industry, costing an estimated five to eight percent of revenues per annum.

Fraud detection can predict and prevent fraud. Newer cases, such as the types of fraud we already know, are detected using guided learning algorithms that provide a list of the most suspected cases. These cases will be further investigated to see if they are indeed fraudulent.

New types of fraud can also be automatically detected from a database using unsupervised learning algorithms. They can be used to identify extreme, unusual cases that deviate from the average behaviour patterns. These cases are flagged as suspicious and should be investigated further.

Using the above two methods simultaneously, both known and new types of suspicious fraud can be quickly identified and referred to the special investigation team.