Training an algorithm to become a world-class fake finger-detective – interview with Tim Swyka

interview with Tim-Swyka-product manager precise-biometrics

Machine learning (AI) is a key success factor behind Precise Biometrics spoof and liveness detection solution. Tim Swyka, Product Manager at Precise Biometrics US office in Potsdam, NY, is responsible for overseeing this suite of products.

“To get the best results for our customers, we work with them from the initial vulnerability analysis to the final product delivery to provide the best possible experience”, says Tim. “Typically we’ll begin with an analysis for each sensor to determine its vulnerability to spoof attacks. Once we’ve determined the vulnerability profile for the sensor, we’ll begin the onboarding process to customize our solution to fit the vulnerability profile”.

The goal of the onboarding is to train the algorithm to distinguish real fingerprints from spoofs when using the customer’s fingerprint sensor.

Overseeing the Onboarding Process
Among Tim’s responsibilities is to coordinate the liveness detection onboarding process for each fingerprint sensor.

”Our solution is machine learning-based, so capturing training data is a vitally important part of the process. We’ll capture training samples from real fingers as well as fake fingers created out of household materials such as gelatine, latex paint, or modelling clay.”

After the collection phase, Tim starts the training phase, running the software and analysing the results.

“Once we have determined that the software works correctly and has learned to tell the difference between the live fingers and the spoofs, we can provide the final configuration to the customer for integration with their sensor”.

Comprehensive Experience with Fingerprint Liveness
Tim is a veteran at the company. With a master’s degree in Electrical Engineering from Clarkson University, Tim started working at the company in 2011. In the past six years, he’s had a variety of different responsibilities focused around liveness detection and spoof mitigation.

”I’m a hardware guy, turned software guy. Most of my background was in hardware design but I’ve really enjoyed working in the software world. Having a deep understanding of hardware is extremely useful when optimizing software for a specific platform.”

When asked what has prepared him most for his current position, Tim says that mastering the basics has been key.

”Software development is fun, but touch typing and communication skills are two things that impact the work I do every single day.”

What is the biggest change in your work so far? 
“Our anti-spoof software has, until recently, been delivered as a stand-alone product. Now it is offered totally integrated with the industry leading fingerprint matching software Precise BioMatch Mobile. This is a powerful and totally unique combination of efficient algorithm based-solutions for fingerprint matching and spoof and liveness detection that has come at the right time”.

What do you do when you’re not at the office?
”I enjoy spending time with my lovely wife and our new baby daughter. In my free-time you’ll usually find me working on projects around the house or volunteering as a sound engineer for our local church.”