South African scientists at the University of Johannesburg’s aiThenticate Computervision Labs have developed what they call a world-first next-generation digital identification technology to fight identity theft.
aiThenticate Computervision Labs is an initiative by the University of Johannesburg to develop disruptive technology innovations. It operates as an independent commercial enterprise.
According to the firm, the technology, dubbed aiDX, is an advanced identification technology that uses deep science to simulate human cognition in order to confirm or determine the identity of a person to industrial grade standards on any smartphone or device equipped with a camera.
“We have been working on this technology for the past five years,” says André L Immelman, CEO of aiThenticate Computervision Labs. “The whole initiative was based on the understanding that one of the most important and difficult questions of our time is ‘who is somebody actually?’ We haven’t figured out how to answer that quite properly. In the past, we have use all sorts of artefacts like passwords, biometrics, etc.
“However, in the post-9/11 world, where the security ante has been raised tremendously, we need to have answers that are certain about identity. We have been working on this solution for the past five years without commercialising it.”
aiDX is set to be commercially launched next month.
The company points out that every working day, an estimated 16.8 billion “authenticated transactions” are effected as someone, somewhere in the world, authorises a payment or a money transfer, a purchase or an investment, among other things.
“aiDX breaks with convention by using artificial intelligence to answer one of the most difficult, most challenging and most urgent questions of our time: ‘who is someone… actually?’ And it does it all on the one device we all carry with us all over all the time – a standard, off-the-shelf smartphone or tablet,” says Immelman.
aiThenticate Computervision Labs says the failure of conventional authentication methods such as signature, identity artefacts, passwords, PINs, etc, to effectively arrest identity theft, has seen a rapid shift towards biometrics as a means of authenticating a person over recent years: fingerprints, face prints, voice prints, iris prints, etc.
However, it points out that last year, the global loss from identity theft was about $2 trillion, and it is doubling every year.
“In SA, R1 billion was lost in SIM card swaps last year. These figures go to show just how ineffective conventional biometrics is in a world where someone sitting at his PC in one country is able to hack into a bank account in another country, even on a completely different continent,” says Immelman.
“Conventional biometrics are based on simple geometry – connecting key features to form a pattern that is then associated with a particular individual – it’s a bit like a child’s game of ‘connect the dots’ to form a picture. However, while conventional biometrics may be sufficient for the purposes of unlocking a smartphone, the scarcity of key features that are generally visible in a latent fingerprint or a face print, for example, means this system tends to fail rapidly with larger population groups.”
Immelman says the simple mathematics that underscores conventional biometrics explains why misidentification is a very real problem with fingerprint, face print, voice print and iris print solutions, rendering conventional biometrics inadequate as a real-world authentication solution.
The disruptive technology innovation promises to completely revolutionise the way people traditionally proceeded when authenticating an individual, he adds.
“Inspired by brain function and using advanced deep learning techniques, research scientists at aiThenticate Computervision Labs have successfully been able to simulate human cognition – what we as humans instinctively do when we recognise someone, even after many years, or even without ever having met someone in person.”
He notes that representing a quantum leap over conventional facial recognition methods, neural cognition yields results that are not only far more reliable and far more robust, but that are in fact impervious to the factors that tend to interfere with conventional facial recognition systems. These include ageing, make-up, facial hair, weight loss/gain or spectacles.