A robot which learns in the same way a young child would has been designed by Norwegian researchers, with no predefined knowledge database to guide its actions.
Two researchers from the Norwegian University of Science and Technology (NTNU) have devised a machine that analyses sound through a system based on the human ear and learns to recognise images using a digital model of how nerve cells handle sensory impressions in the brain.
The robot, called Self, uses deep learning which is loosely inspired by the neural circuitry of the human brain when it perceives and interacts with the world. It learns entirely from sensory input, the researchers said, with no preset knowledge database, resembling the same learning process of a child in real life.
“We’re still pretty far away from accurately modelling all aspects of a living child’s brain, but the algorithms that handle sound and image processing are inspired by biology,” says Øyvind Brandtsegg, a professor at NTNU. “We’ve given it almost no predefined knowledge on purpose.”
The robot ‘hears’ sounds from a person speaking and can connect these to a simultaneous video feed of the speaker. It picks up a sound that the person appears to be emphasising and responds by playing other sounds that it associates with this. The neutral representation of its sound-pictures association shows how the ‘brain’ connects the two.
The robot has already interacted with different audiences, with popular lines such as “My name is…” and “What is your name?” being thrown at it, but also poems and colloquialisms. The robot gradually absorbed more and more impressions of different people, the researchers said.
“Many artificial intelligence (AI) researchers, myself included, believe that true intelligence can’t occur in a vacuum. It is a consequence of adapting and living in a dynamic environment,” Axel Tidemann, co-researcher, said. “You could see our intelligence as a by-product of our adaptability.”
“We believe that the right way to reach for the ‘holy grail’ of AI is to implement biologically inspired models in a machine. Let it operate in a physical environment and see if we can observe intelligent behaviour.”