Connect with us

Artificial Neural Networks

Researchers Mimic Sea Slug Strategies in Quantum Material

Published

 on

Image: Purdue University photo/Kayla Wiles

Researchers at Purdue University have discovered that a material can mimic the most essential intelligence features of a sea slug. This could help them build hardware that would make AI more efficient and reliable, which could impact many fields like autonomous vehicles, surgical robots, and social media algorithms.

The study was published in Proceedings of the National Academy of Sciences. It was led by a team of researchers from Purdue University, the University of Georgia, Rutgers University, and Argonne National Laboratory. 

Shriram Ramanathan is a Purdue professor of materials engineering.

“Through studying sea slugs, neuroscientists discovered the hallmarks of intelligence that are fundamental to any organism’s survival,” said Ramanathan. “We want to take advantage of that mature intelligence in animals to accelerate the development of AI.”

Learning From Sea Slugs

Neuroscientists have learned two main signs of intelligence from sea slugs: habituation and sensitization. Habituation takes place when something or someone gets used to a stimulus over time, while sensitization is when something or someone reacts strongly to a new stimulus.

One of the major problems with AI is the “stability-plasticity dilemma,” which takes place when AI has trouble learning and storing new information without overwriting existing information. With habituation, AI could “forget” unneeded information to become more stable. At the same time, sensitization could help it retain new and important information, which enables plasticity.

Demonstrating the Strategies in Nickel Oxide 

The researchers were able to demonstrate both habituation and sensitization in nickel oxide, which is a quantum material due to its properties that can’t be explained by classical physics. If a quantum material could successfully use these forms of learning, it could be possible to build AI directly into hardware. The AI could perform more complex tasks while using less energy if it could operate through both hardware and software.

“We basically emulated experiments done on sea slugs in quantum materials toward understanding how these materials can be of interest for AI,” Ramanathan said.

Sea slugs demonstrate habituation when they stop withdrawing their gills as much, which is a response to being trapped on the siphon. An electric shock to the tail of a sea slug causes the gills to withdraw more dramatically, which is sensitization. 

In order to reproduce this in nickel oxide, there must be an increased change in electrical resistance. By repeatedly exposing the material to hydrogen gas, nickel oxide’s change in electrical resistance decreases over time. However, when a new stimulus such as ozone is introduced, the change in electrical resistance greatly increases. 

The research group took these findings into account, and a team led by Kaushik Roy, Purdue’s Edward G. Tiedemann Jr. Distinguished Professor of Electrical and Computer Engineering, modeled nickel oxide’s behavior. They built an algorithm that could use the habituation and sensitization strategies to categorize data points into clusters.

“The stability-plasticity dilemma is not solved at all. But we’ve shown a way to address it based on behavior we’ve observed in a quantum material,” Roy said. “If we could turn a material that learns like this into hardware in the future, then AI could perform tasks much more efficiently.”

In order for this to be useful and practical, the researchers must find out how to apply these strategies in large-scale systems, as well as determine how a material could respond to stimuli while it’s integrated into a computer chip. 

Alex McFarland is a historian and journalist covering the newest developments in artificial intelligence.