22-06-2018 | | By Rob Coppinger
Neuromorphic computing, electronic devices that imitate the processes of the human brain for faster, more energy efficient, computation could have come one step closer with the development of a nanoscale device that acts like a neuron.
Neurons in the brain are the fundamental unit that makes it work, sending signals among the 100 billion neurons that power a person’s intellect. The neuron can have different electrical states, which enable it to send signals. For this reason, a memristor has been compared to a neuron, but a memristor does not transmit anything. The key advance in the nanoscale neuromorphic device is its transmission of a soliton wave. This is a magnetic wave that only travels through magnetic material. While the general idea of electromagnetism is that it propagates everywhere, for example, the way a magnet attracts iron filings, a soliton does not travel beyond the magnetic material.
“Neurons can have a different level of charge…and they communicate in some way by axons,” said New York University physicist, Andrew Kent, who led the soliton research. An axon is one of three main parts that make up a neuron. He added: “In the electronics community it’s been interesting to try to find analogues to the way the brain might compute.” His team developed the 10-nanometre diameter device that generates soliton waves. This soliton device can retain a charged state, send a soliton, compute and keep track of “where it has been in its computational space.”
Soliton waves would transmit the signal through a circuit of magnetic material in a far more energy efficient way than today’s current carrying circuits. Kent estimates that each soliton device would use 10-100 microwatts. The human brain’s total power consumption is only a few Watts and it has 100 billion neurons. However, that low power consumption is reflected in the fact that the brain operates at the speed of Hertz, while the soliton’s speed is measured in Gigahertz. “If we can figure out something the brain does, we can do it much faster. But, the brain has enormous connectivity which compensates for its low speed, and it has very low power consumption,” Kent explained.
The amplitude of the soliton is shown as a function of its position. The left image is a plot of amplitude versus its position and the right image represents the same object on a colour scale.
While this soliton device is analogous to a neuron, today’s neural networks are not the same thing. They are only software representations of analogous brain-like functions, emulating how the mind works. “I think what we’ll see is the emulation will get so far, but to go further you will need hardware and the question is, what hardware.” The network is only the digital ones and zeros that flow around a conventional semiconductor microchip. Neuromorphic computing is the electronic equivalent of a human brain, with comparable components and structure.
“What we’re hoping in the magnetics community is developing devices that can, themselves, do the computation and that will be much faster than emulating it on a digital machine,” said Kent. “We’re not building a neuromorphic circuit, we’re just showing one component in the scheme of things.” He does not think they have, “exactly the right device yet, but we have lots of people working on this”. Once a neuromorphic machine is developed, Kent thinks, “it could be a collection of devices and magnetic [devices] might be one part of the whole scheme.”
The research team included academics from the University of Barcelona, the Institute of Materials Science of Barcelona and Physikalisch-Technische Bundesanstalt in Germany.