Developers from Purdue University created a neural network that can analyze fMRI brain images taken while watching a video, and then determine in real time what exactly the person was watching.
The experiment involved three subjects, which showed about a thousand small videos. During the show, scientists were able to obtain a huge amount of fMRI data, which then began to demonstrate a convolutional neural network, trained to compare brain activity with the plots of video clips. According to the fMRI snapshot, the network learned very quickly and correctly to determine what exactly the volunteer was watching at the time when the picture was taken.
In addition, the neural network learned to decipher data from other people based on information obtained from fMRI of other volunteers, and the result was equally high with both fMRI data from healthy subjects and those with visual defects.
Thanks to this study, scientists were able to decipher the thoughts, and at the same time they found out which parts of the brain are responsible for the recognition of images and video clips. The fact is that the brain divides the video into separate components. For example, if a person sees a car moving against the wall, one brain area recognizes the wall, and another car - in this way, scientists managed to trace the brain's work when comparing individual blocks of information and putting it into a single picture.
The article is based on materials .
Comments
Post a Comment