Researchers have designed a latest artificial intelligence network that can interpret what a person is viewing and decode the human mind by analyzing brain scans. The improvement might assist efforts to enhance AI (artificial intelligence) and result in new approaches into the function of the brain. Critical to the study is a kind of algorithm dubbed as a convolutional neural system, which has been involved in allowing smartphones and computers to recognize objects and faces. “That type of system has made a huge effect in recent years in the field of computer vision,” claimed an assistant professor in the U.S. at Purdue University, Zhongming Liu, to the media in an interview.
“Our system utilizes the neural network to know what you are viewing,” Liu claimed. Convolutional neural networks, a type of algorithm related to deep learning, have been utilized to survey how the brain executes visual stimuli and other static images. “This is the initial time such a method has utilized to witness how the brain processes natural scenes of movies—a step toward understanding the brain while users are making to make sense of dynamic and complex visual surroundings,” claimed a doctoral student at Purdue University, Haiguang Wen, to the media in an interview last week. Wen is also the lead author of the study posted in the journal Cerebral Cortex. The scientists gained 11.5 Hours of fMRI (Functional magnetic resonance imaging) data from each of 3 female subjects seeing 972 video clips, comprising those displaying animals or people in nature or action scenes.
The information was utilized to train the network to forecast the activity in the visual cortex of the brain while the females were seeing the clips. The system was then utilized to understand fMRI information from the females to reconstruct the clips, even the ones that the system had never seen before. The system was capable of accurately decoding the fMRI information into particular image segments. Actual images and video were then represented side-by-side with the interpretation of the computer as to what the brain of the person saw on the basis of fMRI information.