Faces are very important to us. Our brains can recognise a face in milliseconds, even if we retain no other information about that person. Recording and recalling faces is so intuitive that there is even a condition called prosopagnosia, which occurs when someone struggles to remember faces. Not surprisingly, scientists have discovered our brains have areas dedicated solely to managing information about faces.
Humans are very social creatures, and we rely on quickly identifying friends from foes in a group of faces. Ever since we invented computers, we’ve aimed to endow them with the same abilities. Today, we have computers that understand human speech, translate languages, and recognise our faces. Facial recognition biometrics are today more accurate than the human eye can ever be.
Facial recognition is already used extensively to identify individuals, help prevent crimes such as fraud and make it easier for people to authenticate and access services through their digital identity. We already unlock our phones with a glance many times a day. Imagine if we could pop into the doctor’s office or breeze through airports using our faces to unlock everything, safely and securely.
Until recently, there have been two key limitations around facial recognition. The first is scale. Finding a face amongst thousands, millions, or billions without any reference, like an ID, to narrow the search is like looking for a needle in a haystack. It’s very intensive and time-consuming, sometimes more so than what makes the process worthwhile.
Then there are important privacy concerns. For such a system to work fast, it often relies on other personal data, such as your ID, email address or mobile number. If someone were to access that data used alongside your photo, they could learn a lot about you.
Exciting new developments in this space are resolving these issues. Specifically, a technology called CryptoNets, using fully homomorphic encryption, is changing how fast and securely computers can identify a face. (Homomorphic Encryption refers to a new type of encryption technology that allows computation to be directly on encrypted data, without requiring any decryption in the process.)
CryptoNets uses artificial intelligence and a new type of encryption to deliver on two fronts. It is very fast, locating a face among countless others in milliseconds without the need to have any other data to narrow the search. It is also highly efficient, requiring far fewer computing resources than other types of facial recognition when making one-to-many comparisons or the traditional one-to-one face comparisons.
The technology is also great for privacy. Information encrypted through CryptoNets becomes anonymised data and no longer contains personal information. It is useless without the system, which itself doesn’t need to decrypt the information to use it. So, there is no point in stealing that personal information. Notably, CryptoNets technology converts material such as photos or voiceprints into a different data format that cannot be used outside the CryptoNets context.
The result is a new biometric technology that is much faster, far more secure, naturally compliant with personal information laws such as PoPIA and GDPR, and applicable to different forms of recognition. CryptoNets technology also supports voice, fingerprints and so on. And because of how it works, CryptoNets can quickly identify someone using only one element, such as a photo (no need to give your ID number every time you verify yourself), and does so in near real-time, so it works even in large crowds.
Many of us become nervous when we think of facial recognition. There is always scope for abuse and mistakes. But facial recognition is not going away – it’s just too helpful. So, we should instead ask how we can make it more responsible and respectful of our privacy.
CryptoNets technology achieves privacy and efficiency goals with incredible results. As it grows more popular, it will improve biometric recognition of all types while keeping our personal information safer. It aligns nicely with a wave of improvements in the sector that promotes better ethics and greater accuracy.
Facial recognition is controversial for many reasons. These are being resolved through breakthroughs such as CryptoNets. From fraud prevention to opening your front door with your voice, it will improve many aspects of our lives in fantastic ways without compromising our safety and privacy.
Digital identity orchestration company, Contactable, will be the first organisation in Africa to pioneer this technology as part of its extended product offerings in the market.
© Technews Publishing (Pty) Ltd. | All Rights Reserved.