This Quantum material is learning like a living organism

Researchers at Rutgers University in the United States have discovered that learning, a universal characteristic of intelligence in living things, can be mimicked in quantum matter, as reported in a release.

A quantum material is one whose properties cannot be described by classical physics, but by quantum physics. Mott insulators are a class of quantum materials.

Mott insulators exhibit extraordinary phenomena, such as high-temperature superconductivity and the so-called colossal magnetoresistance (which allows them to considerably change their electrical resistance in the presence of a magnetic field), due to the interaction between charge and spin.

What the new research has found is that Mott’s insulators can “learn” to respond to external stimuli in a way that mimics animal behavior. The results of this study are published in PNAS.

Quantum material

Quantum material The new research has focused on the nickel oxide (NiO), a metal oxide that is a clear example of Mott’s insulating material: he studied how electrical conductivity, a special type of insulating material, responds when its environment changes repeatedly over various time intervals.

Learning can be mimicked in synthetic matter, a discovery that in turn could inspire new algorithms for artificial intelligence (AI). (Courtesy: Rutgers University-New Brunswick)

And discovered that this quantum material mimics associative learning: by means of oxygen, ozone and light stimuli, it changes its electronic properties based on what it has learned.

Associative learning is one of the basic ways that living organisms can learn and, in this way, constantly adapt to changes.

Another intriguing finding from the study is that when the researchers exposed nickel oxide to rapidly changing oxygen concentrations or different intensities of light, the material barely reacted, instead remaining in an unstable state with little fluctuation in electrical conductivity, stand out PhysicsWorld.

Universal learning

Universal learning However, when scientists introduced additional atomic defects, using a more severe stimulus, such as ozone, the material’s electrical conductivity fluctuated faster, only to slow down again.

Subhasish Mandal, one of the researchers, highlights that his discovery shows that there are universal learning characteristics, such as habituation and sensitization, which are generally found only in living species, in quantum materials.

The researchers go even further: they think that this learning capacity can be part of other quantum materials and be used to modulate their properties based on certain external stimuli.

Revolution in AI

Revolution in AI The discovery of behaviors such as habituation and sensitization in these non-living materials could lead to new algorithms for artificial intelligence (AI), the researchers say.

Emulating this behavior, typical of living organisms, in an inert material that is in a solid state, could inspire new algorithms in artificial intelligence and neuromorphic computing, according to the scientists.

Those new algorithms would have the flexibility to deal with uncertainties, contradictions, and other aspects of everyday life, just as they do at the complex levels of brain and nervous system life.

Neuromorphic computing mimics the neural structure and function of the human brain, in part, by building artificial nervous systems to transfer electrical signals that mimic brain signals.

Autonomous AI

Autonomous AI Each of the 100 billion neurons in the human brain, for example, receives electrical inputs from some of its neighbors and then “fires” an electrical output to others when the sum of the inputs exceeds a certain threshold.

This process, also known as “spiking,” can be replicated in nanoscale devices such as spintronic oscillators, PhysicsWorld emphasizes.

In addition to being potentially much faster and energy efficient than conventional computers, devices based on these neuromorphic principles, thanks to the new discovery, could learn to perform new tasks without being directly programmed to perform them, the researchers conclude.

This research, therefore, will considerably boost the possibilities of AI, a technology that can not only think, reason, act and behave like people, and even imagine the invisible, but now you can also learn how a brain does it.

Reference:

Neuromorphic learning with Mott insulator NiO. Zhen Zhang et al. PNAS September 28, 2021 118 (39) e2017239118. DOI:https://doi.org/10.1073/pnas.2017239118