Fraud Investigators: The Secret Component to Help ML Models Detect Synthetic Identities

As the world moves toward even greater levels of digital commerce, and fraudsters evolve from merely impersonating other people to fabricating fake (aka “synthetic”) identities—a particularly dangerous type of identity fraud—automated fraud detection is the only way to keep up with scaled attacks.

Human-in-the-loop vastly improves the performance and transparency of machine learning (ML) datasets for automated models to detect synthetic identity fraud. Fraud investigators uncover patterns and apply domain knowledge to identity fraud labels to train the ML platform. This means synthetic identities are detected with far greater precision, which dramatically reduces false positives and manual reviews.

Check out our on-demand webinar to hear from a front-line fraud investigator why synthetic identity fraud is so complex and difficult to detect, and learn about the behaviors and patterns that set it apart. Take part in an interactive opportunity to test your skills at choosing between synthetic and legitimate identity profiles.

Key takeaways:

• Learn about human-in-the-loop machine learning and the role that fraud prevention experts are playing in using domain knowledge to eradicate synthetic identity fraud.
• See some of the particular patterns that synthetic identities exhibit.
• Guess the profile of the most common synthetic identity by name, gender, age, location, birth month, and more.
• Discern how to prevent a simple typo from pegging you as a synthetic identity.
• Try your hand at separating synthetic identity profiles from legitimate consumers.

Bre Reimer

Senior Fraud Investigator, Socure

Mike Cook

VP, Commercialization - Fraud Solutions, Socure