International research team develops new hardware for neuromorphic computing

In the future, modern machines should not only follow algorithms quickly and accurately, but also work intelligently, that is, in a way that resembles the human brain. Scientists from Dortmund, Loughborough, kyiv and Nottingham have developed a vision-inspired concept that could make future artificial intelligence much more compact and efficient.

They built an on-chip phonon-magnon tank for neuromorphic computing that was recently touted as the publisher’s highlight by Natural communications.

Human sensory organs convert information such as light or smell into a signal that the brain processes through myriad neurons connected by even more synapses. The brain’s ability to train itself, namely to transform synapses, combined with the large number of neurons, allows humans to process and respond quickly to very complex external signals.

Researchers are attempting to mimic the principle of signal transmission and entrainment with complex neuromorphic computing systems, systems that resemble the neurobiological structures of the human nervous system. However, modern technologies are still infinitely far from achieving comparable information density and efficiency.

One approach to improve neuromorphic systems is the reservoir computational framework. Here, input signals are mapped into a multidimensional space called a reservoir. The reservoir is not trained and only accelerates recognition by a simplified artificial neural network.

This results in a huge reduction in IT resources and training time. Human vision is a typical example of natural reservoir computing: in the eye, visual information is preprocessed by hundreds of millions of photoreceptors in the retina and converted into electrical signals that are transmitted to the brain via the optic nerve. This process significantly reduces the amount of data processed in the brain by the visual cortex.

Modern computer systems can emulate tank functions when processing digitized signals. However, a fundamental breakthrough will be made when reservoir calculation can be performed directly with analog signals by a natural physical system, as in human vision.

The international team consisting of researchers from Dortmund, Loughborough, kyiv and Nottingham has developed an innovative concept that brings such advances significantly closer.

The concept suggests a reservoir based on acoustic waves (phonons) and spin waves (magnons) mixed in a 25x100x1 cubic micron chip. The chip consists of a multi-mode acoustic waveguide through which many different acoustic waves can be transmitted and which is covered with a 0.1 micron thick patterned magnetic film.

The information delivered by the train of ultrashort laser pulses is preprocessed before recognition by conversion into a propagating phonon-magnon wave packet. The short wavelengths of the propagating waves result in high information density, enabling reliable recognition of visual shapes drawn by a laser over a remarkably small area of ​​less than one photopixel.

Professor Alexander Balanov of Loughborough University, one of the authors of the concept, says: “The potential of the proposed physical system as a reservoir was immediately evident to us due to its astonishing combination of variability and multidimensionality. »

His colleague, Professor Sergei Savel’ev, emphasizes the similarity of the demonstrated principle of operation with the functionality of the human brain: “The functionality of the developed tank is based on the interference and mixing of optically generated waves, which is very similar to this recently suggested mechanism of information processing in the biological cortex.

Dr. Alexey Scherbakov, who led the project at TU University Dortmund, says: “Our concept is very promising because it is based on the conversion of the income signal into high-frequency acoustic waves, as in wireless communication devices. modern threads. »

“Our acoustic frequency range above 10 GHz is a bit higher than currently available, but it is targeted by upcoming wireless communication standards. So, who knows, probably in a few years your cell phone will help you to make very human decisions.”