Researchers at the University of Sydney and Japan’s National Institute for Material Science (NIMS) have actually found that an artificial network of nanowires can be tuned to respond in a brain-like way when electrically promoted.
The global group, led by Joel Hochstetter with Teacher Zdenka Kuncic and Professor Tomonobu Nakayama, discovered that by keeping the network of nanowires in a brain-like state “at the edge of mayhem,” it carried out tasks at an optimal level.
This, they say, suggests the hidden nature of neural intelligence is physical, and their discovery opens an exciting avenue for the development of artificial intelligence.
The study is released today in Nature Communications.
“We utilized wires 10 micrometres long and no thicker than 500 nanometres arranged randomly on a two-dimensional airplane,” stated lead author Joel Hochstetter, a doctoral candidate in the University of Sydney Nano Institute and School of Physics.
“Where the wires overlap, they form an electrochemical junction, like the synapses in between nerve cells,” he stated. “We discovered that electrical signals executed this network immediately discover the best route for transferring info. And this architecture allows the network to ‘keep in mind’ previous paths through the system.”
ON THE EDGE OF MAYHEM
Utilizing simulations, the research study team evaluated the random nanowire network to see how to make it finest perform to solve simple tasks.
If the signal promoting the network was too low, then the paths were too predictable and orderly and did not produce complex adequate outputs to be useful. If the electrical signal overwhelmed the network, the output was totally chaotic and ineffective for issue solving.
The optimal signal for producing a helpful output was at the edge of this disorderly state.
“Some theories in neuroscience suggest the human mind could operate at this edge of turmoil, or what is called the critical state,” stated Professor Kuncic from the University of Sydney. “Some neuroscientists think it remains in this state where we achieve optimum brain efficiency.”
Professor Kuncic is Mr Hochstetter’s PhD consultant and is presently a Fulbright Scholar at the University of California in Los Angeles, working at the intersection between nanoscience and artificial intelligence.
She said: “What’s so amazing about this result is that it suggests that these kinds of nanowire networks can be tuned into regimes with diverse, brain-like cumulative characteristics, which can be leveraged to optimise info processing.”
GETTING RID OF COMPUTER SYSTEM DUALITY
In the nanowire network the junctions between the wires enable the system to integrate memory and operations into a single system. This differs from basic computers, which different memory (RAM) and operations (CPUs).
“These junctions imitate computer system transistors but with the extra property of remembering that signals have taken a trip that pathway before. As such, they are called ‘memristors’,” Mr Hochstetter said.
This memory takes a physical form, where the junctions at the crossing points between nanowires imitate switches, whose behaviour depends upon historic response to electrical signals. When signals are applied across these junctions, tiny silver filaments grow activating the junctions by enabling current to flow through.
“This develops a memory network within the random system of nanowires,” he stated.
Mr Hochstetter and his group built a simulation of the physical network to demonstrate how it might be trained to solve really simple jobs.
“For this research study we trained the network to transform a simple waveform into more complex kinds of waveforms,” Mr Hochstetter said.
In the simulation they changed the amplitude and frequency of the electrical signal to see where the very best performance occurred.
“We discovered that if you press the signal too gradually the network simply does the same thing over and over without learning and developing. If we pushed it too set, the network ends up being erratic and unforeseeable,” he stated.
The University of Sydney scientists are working closely with collaborators at the International Center for Materials Nanoarchictectonics at NIMS in Japan and UCLA where Teacher Kuncic is a visiting Fulbright Scholar. The nanowire systems were established at NIMS and UCLA and Mr Hochstetter developed the analysis, dealing with co-authors and fellow doctoral trainees, Ruomin Zhu and Alon Loeffler.
REDUCING ENERGY CONSUMPTION
Teacher Kuncic stated that joining memory and operations has big practical advantages for the future development of expert system.
“Algorithms needed to train the network to know which junction should be accorded the appropriate ‘load’ or weight of information chew up a great deal of power,” she stated.
“The systems we are establishing do away with the need for such algorithms. We just permit the network to establish its own weighting, suggesting we only require to stress over signal in and indicate out, a structure known as ‘reservoir computing’. The network weights are self-adaptive, potentially freeing up large amounts of energy.”
This, she stated, indicates any future artificial intelligence systems using such networks would have much lower energy footprints.