You have probably heard the word biometrics several times in recent times. The digital transformation, accelerated in many areas by the pandemic, has sparked a broad social debate in many areas, including biometrics and artificial intelligence, but what exactly do we mean when we talk about biometrics? Want to know more about biometrics? Click here: [ Ссылка ]

To understand the origin of biometrics we must go back to about 5 or 7 years ago, when biometric models were based on Landmarks, this method involves taking measurements between multiple points, for example of our face, obtaining, as a result, a vector that is a summary of all those measurements. This is what we are probably all familiar with. It has some loopholes, so to speak, at the level of data protection.

And it is that in view of that vector and knowing more or less how it has been constructed, we can interpret the measurements it contains and therefore re-represent the original image from which it started.

In addition, in terms of accuracy, it has very low accuracy guarantees, it would not reach 95%, however, today the state-of-the-art biometric models are based on artificial intelligence and more specifically, on neural networks.

This has some key features that, as we said before, make them secure by default. The result of these models is that we are going to get a mathematical vector that depends on the artificial intelligence of that particular biometric engine.

This has some consequences, firstly, that the resulting vectors of these models based on artificial intelligence will not be able to be reversed to obtain the image from which it was started, the biometric characteristic. That is, we are not going to be able to arrive at the exact image, for example, of our face that we used to generate our vector.

With these biometric models, therefore, we could say that this vector is irreversible and private. Also, as we have said, the vector is going to depend on each biometric engine with which it is created.

Therefore, if we pass, for example, a facial image through two different biometric engines or even through a different version of the same biometric engine.
The vector that we will obtain will be different or, in other words, each vector will only be interpreted a posteriori by the exact version of the biometric engine that created it.

In other words, a vector created with Veridas technology cannot be used for any other system in the world and is therefore not interoperable in terms of data protection, as this gives us many guarantees.

свернуть