Neuromorphic computing: Phonon-magnon reservoir, eye-like metasurface, graphene/water memristive device.
Researchers from TU Dortmund, Loughborough University, V. E. Lashkaryov Institute of Semiconductor Physics, and University of Nottingham were inspired by the human eye to propose an on-chip phonon-magnon reservoir for neuromorphic computing.
In reservoir computing, input signals are mapped into a multidimensional space, which is not trained and only expedites recognition by a simplified artificial neural network. The eye is an example of natural reservoir computing, in which the retina’s photoreceptors pre-process information and convert it into electrical signals, reducing the amount of data processed in the brain by the visual cortex.
The team investigated a reservoir based on acoustic waves (phonons) and spin waves (magnons) mixed in a chip of 25x100x1 cubic microns. The chip consists of a multimode acoustic waveguide through which many different acoustic waves can be transmitted and which is covered by a patterned 0.1-micron-thickness magnetic film. The information delivered by the train of ultrashort laser pulses is pre-processed prior to the recognition by conversion to the propagating phonon-magnon wavepacket. Short wavelengths of the propagating waves results in high information density, which enables the confident recognition of visual shapes drawn by a laser on an area of less than one photopixel.
“The functionality of the developed reservoir is based on the interference and mixture of the optically generated waves, which is very similar to the recently suggested mechanism of the information processing in the biological cortex,” said Sergey Savel’ev, a professor at Loughborough University, in a statement. [1]
Researchers from Pennsylvania State University created a metasurface to mimic the instantaneous image processing power of the human eye. The metasurface can be used to preprocess and transform images before they are captured by a camera, reducing the power and bandwidth required by the computing system.
The metasurface works by converting an image from the Cartesian coordinate system, where image pixels are arranged in straight rows and columns along the x and y axes, to the log-polar system, which uses a bullseye-like pixel distribution. Since it works using nanostructures that bend light, the metasurface does not need any power and works at the speed of light.
“Like the arrangement of light receptors inside the human eye, the metasurface takes images and arranges them in a log-polar coordinate system — with denser pixels for the central, focused features and sparser pixels for the peripheral regions,” said Xingjie Ni, associate professor of electrical engineering and computer science at Penn State, in a release. “This allows for the more important aspects of a photo to come through clearly while others remain less in focus, thereby saving data bandwidth.”
By placing a different metasurface in front of a camera, researchers also can transform the log-polar image back into the original image with Cartesian coordinates. [2]
Researchers from the Max Planck Institute for Polymer Research and Southeast University designed an artificial neuron based on graphene and water in an effort to realize memristive behavior in a relatively simple system.
The aqueous proton-based memristive device is based on a calcium fluoride (CaF2)-supported monolayer graphene in contact with bulk water. The memristive behavior arises from the fast proton transfer across the graphene and the relatively slow diffusion process of protons. The researchers said that the device showed long-term and tunable memory (from 60 seconds to 6000 seconds) and potential for large-scale integration and multiplication. [3]
[1] Yaremkevich, D.D., Scherbakov, A.V., De Clerk, L. et al. On-chip phonon-magnon reservoir for neuromorphic computing. Nat Commun 14, 8296 (2023). https://doi.org/10.1038/s41467-023-43891-y
[2] Zhang, X., Zhang, X., Duan, Y. et al. All-optical geometric image transformations enabled by ultrathin metasurfaces. Nat Commun 14, 8374 (2023). https://doi.org/10.1038/s41467-023-43981-x
[3] Yongkang Wang, Takakazu Seki, Paschalis Gkoupidenis, Yunfei Chen, Yuki Nagata, and Mischa Bonn, Aqueous Chemimemristor Based on Proton-permeable Graphene Membranes. Proceedings of the National Academy of Sciences (PNAS), 121 (6) e2314347121 https://dx.doi.org/10.1073/pnas.2314347121
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
Leave a Reply