The science, metrics behind ID verification technology, defined

January 16, 2019

Achieving the goal of beyond-human levels of accuracy in ID verification technology

We understand that in today’s digital economy, most businesses need to improve accuracy when it comes to ID document counterfeit detection. The Identity verification solution a business chooses should have confidence in their science and automated process, and in their expert agent review team’s human intelligence to label datasets and develop insights. The overarching goal of a sophisticated IDV solution is to provide beyond-human levels of accuracy using analytics, data science and machine learning. 

Identity verification metrics, defined

As the science behind fully-automated authentication methods become more complex and vigorous, it’s not hard to feel confused by the technology. Many technical factors go into creating and improving a continuous learning identity verification engine and the search for the right digital identity verification solution to meet the needs of your business can seem daunting. Here are some definitions of metrics to help you better comprehend the identity proofing industry more clearly:


The right approach to address the specific needs of your business

The relationship between false negative and false positive rates is an inversely relative one; it’s a give-and-take approach that can be fine-tuned to fit a business’ needs and the assessment of opportunity costs surrounding false negatives (halting legitimate prospects for further authentication) and false positives (letting forgeries through). In short, as more non-forgeries undergo more friction in order to be authenticated, more forgeries will be stopped. Contrariwise, letting more “good guys” pass through without any friction comes with the possibility that more “bad guys” will pass through. Managing this relationship according to the needs of any one business is not as simple as it may sound – and it’s important to note that more consumers are willing to undergo “positive friction” in order to ensure a safer experience. With data breaches growing more rampant, even millennial and younger consumers have become sensitive to the way in which their personal information will be handled by websites and other platforms through which they conduct business. 

In fact, we can now view this type of “positive friction” as not only a promise to customers that they’re receiving a secure experience and the safety that they value, but also as an investment in research and development. A robust team of PhD and MS scientists, with expertise in computer vision, data science and deep learning, have built and continuously work to improve Mitek’s machine-learning technology, which is informed by large datasets of forged and real documents, always updated with every release.