George Brostoff, CEO, SensibleVision
A couple of decades ago, to deploy an effective AI instance you needed a box the size of a large refrigerator. Deep Blue, the IBM supercomputer that beat the world chess champion Garry Kasparov in 1997 was huge! Even as recently as 2011, IBM Watson required 750 Power servers sitting in a chilled room at the T.J. Watson Research Center to beat the two Jeopardy Grand Champions.
But here we are closing in on the end of 2018 and things are very different. The power of AI, neural networks and deep learning continues to move away from the cloud and out to the edge. As a result, the power of these technologies can be delivered in previously unheard of ways. Most exciting to me is how they are being applied to deliver fast, frictionless and accurate mobile face authentication.
Andrew Ng, who has been involved in AI for many years as both a Stanford professor and as former Chief Scientist at Baidu said, “AI is the new electricity. It will impact and infiltrate every single business.”I completely agree. And this is certainly true when it comes to mobile, 3D face authentication.
An exciting transformation is taking place. We are seeing a major reinvention of mobile security based on the capabilities that AI and machine learning bring to this space. The development of Continuous Security solutions leveraging constant monitoring and validation of end users with 3D face recognition is ushering in a new era of secure authentication.
An increasing number of portable, rugged, trustworthy face-based authentication solutions are becoming available on smaller form factors. Once bulky hardware required to run these powerful algorithms is now available in a pocket-sized rectangle of glass and aluminium. Smartphones, tablets and laptops are all becoming AI-enabled. And with the advent of smaller and smaller 3D cameras, we’ll probably see it in watches and other wearables sooner than later.
Many of these solutions are not yet using true 3D face recognition, but rather 2D solutions that are subject to false recognition) or even fail to recognize people with dark skin. However, the low cost and greater availability of these 3D cameras permit a bright future with blazing fast and secure recognition in all lighting conditions, even of people with the darkest of complexions.
Collecting and managing the tens of thousands of 3D depth data points required to deliver accurate face verification is a task perfectly suited to today’s mobile AI instances. The increasing power delivered by these algorithms is expanding the potential application of AI-driven face authentication beyond narrow government applications or fun low-security applications –- to other industries like financial services, healthcare, and marketing.
Today, banks and other financial services organizations are starting to leverage deep learning models driving face authentication solutions to improve fraud detection, reduce the need for traditional passwords, and improve the ability to quickly and accurately validate a user based on a scan of his or her face. For example, continuous security tracks the authorized user all the time, not just at logins like passwords or fingerprints.
Healthcare providers are using AI and machine learning combined with computer vision to more accurately track everything from medication consumption to pain management regimens.
Although fraught with ethical considerations, marketing is a growing area of facial recognition and AI innovation and one where we can expect to see more adoption. In most parts of the world, people are getting comfortable receiving targeted information about products and services based on a simple look at the camera on their smartphone. An important key for business success when using AI-driven face authentication is to make sure the users have the right to opt in the program and have full control of their biometric information.
There are broad implications for other sectors as well. Take hospitality for example. AI-driven facial recognition is currently being used in hotels in Kunming, China and is part of a larger trend in that country. Even in higher education, we are seeing adoption. Numerous universities are using AI-driven face scanners to monitor students in classrooms and gauge whether they seem to be bored, the implication being that the professors might need to make their lessons more engaging.
As neural networks and machine learning algorithms become more powerful and more portable, they are going to continue to drive adoption of face authentication solutions, especially ones that provide Continuous Security. Painless, fast, accurate authentication will deliver value across verticals and use cases, regardless of whether the goal is to protect access to mission-critical Enterprise data or to buy a birthday present for a friend at your local shopping mall.
Delivering face-based authentication to provide business value requires that huge amounts of data are collected and analyzed quickly. Not to mention the fact that there is a high expectation about the level of consistency and accuracy. AI and machine learning can address these challenges and their usage will continue to expand as companies look to deliver fast, accurate, continuous and transparent face authentication solutions.