This makes possible a variety of frauds.
The researchers used a neural network to generate synthetic fingerprints, which are then used as the access key for the biometric identification system.
Thus, it was shown that fake fingerprints can still be made. According to The Guardian, scientists from the University of new York was able to simulate every fifth sample of a biometric identification system.
At that level the probability of error was less than one in a thousand. To cheat the system, the researchers used two of its nuances. First, from the ergonomic considerations, the majority of devices of the fingerprint reader and don’t read the whole fingerprint at a time.
Instead they remove only the part of your finger that touches the scanner. It is important that such systems are not uniform in all parts of the image to recognize a holistic fingerprint. Instead, they just compare the available bit with what is in the database.
This means that an attacker would need only match one of the hundreds of thousands of samples stored partial fingerprint to gain access. Second, scientists noticed that some of the features of fingerprints more common than others.
This means that the fake, which contains a lot of popular figures, has a high chance to cheat the system.
Based on these observations researchers have created a fake fingerprint that could pass for as many as possible of these. This method is unlikely to help with specific hacking protected biometric data of the account. But when we are talking about a database of prints, it might work.
© 2018, z-news.link. All rights reserved.