By Hamad Ali, European Staff Writer, WatersTechnology
Oracle is using deep learning to find matching patterns for graph analytics within its compliance platform.
Oracle has been developing its Financial Crime and Compliance Studio (FCC Studio) using deep learning and graph analytics, which is the process of analyzing data in a graph format using data points as nodes and relationships as edges. The enhancement looks to allow users to detect repetitive patterns in graphs of data on individuals in order to assist in know-your-customer (KYC) and anti-money laundering (AML) activities.
The automated functionality within the FCC Studio platform looks at scenarios not previously classified as threats in order to find patterns. As analysts provide more feedback, the risk score associated with an individual’s profile over time can then be reduced. “Deep learning is at the heart of what we’re trying to do,” says Frederic Boulier, senior director and global head of financial crime and compliance management (FCCM) solution consulting at Oracle.
Boulier says analysts look at data to find anything that might compromise an individual. If the analyst finds nothing suspicious, the system will learn from the analyst. If the same graphical patterns happen again, the system will dismiss them. This will make onboarding more efficient by reducing the number of times individuals are flagged as suspicious.
Deep learning uses a computer system that mimics the workings of the human brain, called a neural network. Deep neural networks are opaque, but they can process massive amounts of data and can essentially “learn” on their own. Each layer of nodes in a neural network builds on the previous layer—the more layers, the deeper it is. They require massive volumes of data and when at their best, they can find non-linear correlations.
When a bank onboards a new client, KYC/AML specialists need to be able to connect data on this individual with data the bank might already have, as well as external data. For example, Boulier says the data points can be combined with publicly available information, such as the Panama Papers, the 2015 Swiss banking leaks, or last year’s leaks from Mauritius, to provide extra data with which to investigate the individual.
Boulier says there are limitless scenarios in which graph patterns are applicable. “You could be looking at a guy that you’re investigating and find out that this person is connected to a politically-exposed person or to a sanctioned individual, by virtue of some data points,” he says.
You could make this connection if the two individuals have the same street address or IP address. Or perhaps you find that two individuals have the same, highly-complex password, and deduce that those two individuals are actually one person using an alias. This kind of anomaly in the data could signal fraud or AML risk for a bank.
Boulier says Oracle is also working on entity resolution, an approach that allows users to recognize when two data points relate to the same entity, despite appearing to be different. Entity resolution aggregates data from multiple sources to create accurate customer profiles, taking into account errors or slight variations in the data and determining whether or not they relate to the same person or entity.
“If you are mixing two different people who have nothing to do with each other, then you’re basically creating the wrong profile and you’re protecting against the wrong profile,” Boulier says. “So that’s why entity resolution is very important and the quality thereof is very important.”
This article was originally published in January 2020 on waterstechnology.com.
Message me to learn more, or join the conversation by commenting below.
Subscribe to our Blogs:
Oracle Financial Services Blogs: Subscription