Researcher designs a neural network that can distinguish optical illusions
A new “quantum-inspired” AI perception model could be used for pilot training and mental health care.
A researcher at Charles Sturt University has published details of a new form of neural network, based on human perception of optical illusions.
Dr. Ivan Maksymov, senior researcher in physics and machine learning at the Charles Sturt Institute for Artificial Intelligence and Cyber Futures, revealed his work in a paper titled Neural network model of quantum-inspired optical illusions.
Dr. Maksymov’s work has trained a neural network to recognize optical illusions, such as the ambiguous Necker cube. Dr. Maksymov believes his network could be used in various applications, including psychology and psychiatry, as well as in pilot training.
“It can also be used to train pilots, astronauts and drone operators,” Dr. Maksymov said in a statement.
“Optical illusions can significantly affect a pilot’s ability to fly safely, and every pilot should be prepared to handle them.”
While the neural network remains a work in progress, Dr Maksymov believes it could be put to use at the Australian Airline Pilots Academy (AAPA) in Wagga Wagga.
“This same neural network model could be advantageous for psychologists, behavioral scientists, vision scientists, machine learning experts, developers of virtual reality and video game systems, as well as astronaut trainers,” said Dr. Maksymov.
Dr. Maksymov notes that while his methods use quantum processes, he does not suggest that such processes occur in the human brain. However, using “a quantum-like mathematical description” is a useful way to understand how the brain works.
“Collecting data and studying the brain is complex and time-consuming,” said Dr. Maksymov. “For example, I completed this work in two weeks; For a behavioral scientist to do this could take years.
“The ability to conduct this form of research is currently unique and very valuable.”
