The neural network, “a mind reader”, presented in Russia

It is claimed that it is capable of playing the video being viewed.

Scientists representing the Moscow Institute of physics and technology have presented a device that can, in some way, reproduce the thoughts of a person by the electrical activity of his brain. The development can “guess” what image a person sees in front of him.

During the first part of the experiments, the scientists selected ten-second videos related to one of five randomly selected topics: “abstractions,” “waterfalls,” “people’s faces,” “speed”(various races recorded in the first person) and “moving mechanisms.” As the volunteers watched the videos, experts recorded their brain responses using an electroencephalogram. This allowed the specialist to understand how the frequency characteristics of the EEG wave activity vary depending on the category to which the roller belonged.

Subsequently, scientists developed two neural networks, one of which generated images from the “noise,” and the other — created a similar “noise” from the EEG. These neural networks were combined in such a way that together they were able to reproduce the image on the electroencephalogram. Allegedly, the algorithms managed to cope with this task so reliably that in 90 percent of cases it was possible to determine which of the five categories belonged to the video.

Experts suggest that their results bring the creation of devices for post-stroke rehabilitation, controlled by brain signals. The scientific work devoted to the developed neural networks was presented in the online archive of bioRxiv preprints.

Author: Flyn Braun
Graduated from Cambridge University. Previously, he worked in various diferent news media. Currently, it is a columnist of the us news section in the Free News editors.
Function: Editor