Technology

New neurointerface from Facebook reads thoughts with almost no delays

At the moment, technology has already grown tothe moment when computer algorithms together with specific sensors can read brain activity and use it as commands for control. Simply put, it is already possible to create neural interfaces for controlling electronic devices by the power of thought. It sounds fantastic, only if we are talking about complex devices, then until today they had one problem: a big delay in the operation of devices. Why was? Because scientists have managed to develop a neural interface that works almost without delay.

Who created the neural interface

Engineers are behind the development of the neurointerfaceFacebook divisions called Reality Labs and scientists at the University of California at San Francisco. Since 2017, Reality Labs has been engaged in designing and creating interfaces of the brain-computer interface (brain – computer interface - BCI). The potential for technology is enormous. From typing (even without saying it out loud) to control machines and various electronics. And, according to the publication Nature Communications, it was the team of experts who deciphered and translated human thoughts into text.

How the neural interface works

The first prototype of the device appeared in the beginning2019 year. At the same time, the first stages of testing passed. At the beginning of the year, a device from scientists at the University of California could already translate human thoughts into speech. And it turned out he had a very detailed and competent. He had only one “small” flaw: it took several weeks to decode and process the signal. This is where the experience of the guys from Facebook came in handy.

See also: Facebook showed a prototype of a VR bracelet with feedback.

They just had algorithms to work withneural interfaces, but there were no suitable models to optimize your code. The neurointerface itself looks quite simple: on the inner surface of a special helmet there is a series of highly sensitive electrodes that read brain activity. Further, decryption algorithms come into play.

Of course, for the correct operation of the algorithm you needtest it on someone. Scientists recruited a group of people who read brain activity by asking them simple questions like “How do you feel?” Or “What kind of music do you prefer?”. In this case, the answer could not "invent". Certain answers were prepared for each question (24 pieces in total).


Electrodes for reading brain activity

As a result, the algorithm has learned to determine the answers of people.with an accuracy of 61-76% (depending on the answer), which is a very, very good result. Moreover, the delay in recognition turned out to be not significant at all and was measured in several milliseconds. Thanks to the new technology, many people will be able to control with the power of thought almost anything from a computer mouse cursor to a wheelchair or even a military aircraft.

But the algorithm has at the moment and disadvantages. Now it is limited to the data entered in advance. That is, an impromptu response simply does not understand the neural network. However, according to scientists, this is fixable, because to correct this shortcoming, it is enough to integrate self-learning and speech recognition options into the algorithm.

More news you can read in our news Telegram channel.