A shock to the system: using electricity to find materials that can ‘learn’

Scientists seek to create a new generation of supercomputers, drawing inspiration from the most complex and energy-efficient computer ever built: the human brain.

In some of their early forays into building brain-inspired computers, researchers are investigating different non-biological materials whose properties can be matched to show evidence of learning behaviors. These materials could form the basis of devices that can be combined with new software algorithms to enable more powerful, useful, and energy-efficient artificial intelligence (AI).

In a new study by scientists at Purdue University, researchers exposed anoxia-deficient nickel oxide to short circuit electrical impulses and elicited two different electrical responses similar to learning. The result is an all-electric system that exhibits these learning behaviours, said Shriram Ramanathan, a Rutgers University professor. (Ramanathan was a professor at Purdue University at the time of this work.) The research team used resources from the Advanced Photon Source (APS), a facility used in the US Department of Energy’s (DOE) Office of Science at DOE’s Argonne National Laboratory. .

The first response, habituation, occurs when the lightly agitated substance “gets used to”. The scientists noted that although the material’s resistance increases after an initial jolt, it quickly gets used to the electrical stimulation. “Addiction is like what happens when you live near an airport,” said Fanny Rodoulakis, a physicist and beamline scientist at APS. “The day you move in, you think ‘what a racket’, but eventually you don’t even notice it anymore.”

Another response that the substance exhibits, sensitization, occurs when a larger dose of electricity is administered. “With a greater stimulus, the response of the material increases rather than decreases over time,” Rodoulakis said. “It’s like watching a horror movie and then having someone say ‘Boo! From around the corner — you see him really jump.”

“Almost all living things exhibit these two characteristics,” Ramanathan said. “They really are an essential aspect of intelligence.”

Both behaviors are controlled by quantum interactions between electrons that cannot be described by classical physics and that help form the basis of phase transitions in matter. “An example of a phase transition is a liquid becoming a solid,” said Rodoulakis. “The devices we’re looking at are quite on the edge, and the competing interactions that happen at the electronic level can easily be oscillated one way or another by small catalysts.”

Having a system that can be controlled entirely by electrical signals is essential for brain-inspired computing applications, Ramanathan said. “The ability to manipulate materials in this way will allow materials to take over some of the intelligence responsibility,” he explained. “Using quantum properties to embed intelligence in devices is an essential step toward energy-efficient computing.”

The difference between habituation and sensitization may help scientists overcome a challenge in developing AI called the stability-plasticity dilemma. On the one hand, AI algorithms are often very reluctant to adapt to new information. But on the other hand, when they do, they can often forget some of what they have already learned. By creating a habitable substance, scientists can teach it to ignore or forget unnecessary information and thus gain additional stability, while consciousness can train it to remember and integrate new information, thus allowing for flexibility.

“AI often struggles to learn and store new information without overwriting the information already stored,” Rodoulakis said. “Too much stability prevents AI from learning, but too much flexibility can lead to catastrophic forgetting.”

One of the main advantages of the new study was the small size of the nickel oxide device. “This kind of learning has never been done before in today’s generation of electronics without large numbers of transistors,” Rodoulakis said. “This single-junction system is the smallest system yet to exhibit these properties, which has major implications for the potential development of neural circuits.”

To discover the atomic-scale dynamics responsible for habituation and sensitization behaviors, Hua Zhou of Rodolakis and Argonne used X-ray absorption spectroscopy in the 29-ID-D and 33-ID-D beamlines of APS.

An article based on the study was published in the Sept. 19 issue of Mental Health Advanced smart systems.

The research was funded by the Department of Energy’s Office of Science (Office of Basic Energy Sciences), the Army Research Office, the Air Force Office of Scientific Research, and the National Science Foundation.