Replicating learning in electronic devices

Researchers demonstrate learning behavior in a solid-state device controlled entirely by electrical stimuli.

New research has shown that fundamental aspects of learning – a hallmark of intelligence in organisms – can be replicated and electrically controlled in a quantum material.

In doing so, the research could set the stage for smart materials that could form the basis of future computers that take inspiration from the brain and robotic systems that rely on learning.

One of wildlife’s most impressive talents, and key to the survival of many species, is the ability to learn, particularly in response to negative stimuli.

Two important elements of learning are habituation—a reduced response to repetitive stimuli—and sensitization, which is an enhanced response to a noxious stimulus that leads to avoidance.

Awareness, for example, represents the element of learning that prevents a child from touching a hot surface again, even after having experienced the adverse effects of such a stimulus only once. While habituation may explain why we’re cautious about dipping our toes in a warm bath, but not immediately flinching.

Also Read :  Unvaccinated COVID patients 'at greater risk of death' for at least 18 months after infection | Science & Tech News

Learning in electronic devices

Scientists, including researcher Sandip Mondal from the Department of Electrical Engineering at IIT Mumbai, set out to see if these elements of learning could be emulated electronically. “Learning is a characteristic of organisms throughout their lifespan and is closely related to intelligence,” Mondal said. “We wanted to investigate whether it is possible to demonstrate learning behavior in a solid-state device that is controlled entirely by electrical stimuli.”

In an article published in the magazine advanced intelligent systems, The crew demonstrated that it is possible to discover learning properties in carefully prepared devices at room temperature.

“These results are relevant for building brain-inspired computers in the future,” Mondal added. “Learning can be emulated by monitoring the electrical resistance of the device when successive pulses of electrical voltage are applied to it. The continuous change in resistance follows certain trends, such as: B. a slow decrease or a sudden increase, which can be attributed to learning.”

Teach a device to forget

The biggest challenge in designing a device that can respond to stimuli is designing it to respond to electrical stimuli and then “forget” that response once that stimulus ends.

Also Read :  What a Daily Walking Habit Does to Your Body After 50, Says Science — Eat This Not That

To address this, the team turned to a material called nickel oxide (NiO), which has been identified as a promising quantum material because of its unique electrical, magnetic and optical properties. They then identified an array of NiO sandwich layers that could rapidly change their resistance in response to pulses of electrical charge.

This resistance eventually returns to its original value when the pulses stop, but something interesting happens when the pulses resume before the material is fully “reset”.

Mondal and his colleagues found that under these circumstances, the material responds to subsequent pulses of current by reducing its resistive response in response. This mimics the habituation element of learning found in animals. The response can be controlled – increased or decreased – by adjusting the voltage of the charge.

Learning was further demonstrated in the material through the subtle migration of oxygen defects and in the physics discovered in the device under the influence of electric fields.

Also Read :  Regulators Discuss Improving Electronic Communications Services To Consumers

These processes took place at room temperature and just by using controllable electric fields, demonstrating the potential for adaptability under more general circumstances, meaning they could be used in devices that don’t need to operate at extremely low temperatures, for example.

“We were pleasantly surprised to be able to demonstrate both habituation and sensitization in the same material,” said Mondal.

Finally, the researcher explained the difference between studying learning on animals and learning on devices. “While organisms can exhibit complex behavior due to multiple factors, in materials and devices it is possible to isolate the elementary mechanisms that lead to learning,” Mondal said. “So they present us with simpler model systems that we can study.”

Reference: Sandip Mondal, et al, All-electric non-associative learning in nickel oxide, Advanced Intelligent Systems (2022). DOI: 10.1002/aisy.202200069

Feature image: The final post-workout weight patterns with over 60,000 training images. The network achieves an accuracy of 84.8% over the test set of 10,000 images.

Source link