Researchers created illusory motion using deep neural networks, according to a new study published on March 12, 2018.
Deep neural networks (DNNs), is widely used across the field of computer vision and is being vastly used as a tool to study the brain.
Researchers from the National Institute for Basic Biology conducted a study on deep neural networks related to human visual perception of motion. The team trained DNNs with natural scene videos of motion from the point of view of the viewer, and made use of rotating propeller in unlearned videos to verify the ability of the computer to predict motion in the Rotating Snake Illusion. The computer model accurately predicted the magnitude and direction of motion of the rotating propeller in the unlearned videos. They found that this computer model could also identify and predict the rotational motion for illusion images that were not moving physically, similar to human visual perception.
Dr. Watanabe, lead author of the study said, “This research supports the exciting idea that the mechanism assumed by the predictive coding theory is a basis of motion illusion generation. Using sensory illusions as indicators of human perception, deep neural networks are expected to contribute significantly to the development of brain research.”
This study shows how DNN can predict more than physical movements and can be trained to predict other illusory perceptions, which is generally a human phenomenon.