In materia reservoir computing with a fully memristive architecture based on self-organizing nanowire networks

Nature Materials, Journal Year: 2021, Volume and Issue: 21(2), P. 195 - 202

Published: Oct. 4, 2021

Language: Английский

---

Nanoscale neural network using non-linear spin-wave interference

Nature Communications, Journal Year: 2021, Volume and Issue: 12(1)

Published: Nov. 5, 2021

We demonstrate the design of a neural network hardware, where all neuromorphic computing functions, including signal routing and nonlinear activation are performed by spin-wave propagation interference. Weights interconnections realized magnetic-field pattern that is applied on propagating substrate scatters spin waves. The interference scattered waves creates mapping between wave sources detectors. Training equivalent to finding field realizes desired input-output mapping. A custom-built micromagnetic solver, based Pytorch machine learning framework, used inverse-design scatterer. show behavior transitions from linear at high intensities its computational power greatly increases in regime. envision small-scale, compact low-power networks perform their entire function domain.

Language: Английский

---

Reconfigurable training and reservoir computing in an artificial spin-vortex ice via spin-wave fingerprinting

Nature Nanotechnology, Journal Year: 2022, Volume and Issue: 17(5), P. 460 - 469

Published: May 1, 2022

Language: Английский

---

Hands-on reservoir computing: a tutorial for practical implementation

Neuromorphic Computing and Engineering, Journal Year: 2022, Volume and Issue: 2(3), P. 032002 - 032002

Published: July 1, 2022

Abstract This manuscript serves a specific purpose: to give readers from fields such as material science, chemistry, or electronics an overview of implementing reservoir computing (RC) experiment with her/his system. Introductory literature on the topic is rare and vast majority reviews puts forth basics RC taking for granted concepts that may be nontrivial someone unfamiliar machine learning field (see example reference Lukoševičius (2012 Neural Networks: Tricks Trade (Berlin: Springer) pp 659–686). unfortunate considering large pool systems show nonlinear behavior short-term memory harnessed design novel computational paradigms. offers framework circumvents typical problems arise when traditional, fully fledged feedforward neural networks hardware, minimal device-to-device variability control over each unit/neuron connection. Instead, one can use random, untrained where only output layer optimized, example, linear regression. In following, we will highlight potential hardware-based networks, advantages more traditional approaches, obstacles overcome their implementation. Preparing high-dimensional system well-performing task not easy it seems at first sight. We hope this tutorial lower barrier scientists attempting exploit tasks typically carried out in artificial intelligence. A simulation tool accompany paper available online 7 https://github.com/stevenabreu7/handson_reservoir .

Language: Английский

---

In-memory computing with emerging memory devices: Status and outlook

APL Machine Learning, Journal Year: 2023, Volume and Issue: 1(1)

Published: Feb. 14, 2023

In-memory computing (IMC) has emerged as a new paradigm able to alleviate or suppress the memory bottleneck, which is major concern for energy efficiency and latency in modern digital computing. While IMC concept simple promising, details of its implementation cover broad range problems solutions, including various technologies, circuit topologies, programming/processing algorithms. This Perspective aims at providing an orientation map across wide topic IMC. First, technologies will be presented, both conventional complementary metal-oxide-semiconductor-based emerging resistive/memristive devices. Then, architectures considered, describing their aim application. Circuits include popular crosspoint arrays other more advanced structures, such closed-loop ternary content-addressable memory. The same might serve completely different applications, e.g., array can used accelerating matrix-vector multiplication forward propagation neural network outer product backpropagation training. algorithms properties enable diversification functions discussed. Finally, main challenges opportunities presented.

Language: Английский

---

Physical reservoir computing with emerging electronics

Nature Electronics, Journal Year: 2024, Volume and Issue: 7(3), P. 193 - 206

Published: March 12, 2024

Language: Английский

---

Reducing reservoir computer hyperparameter dependence by external timescale tailoring

Neuromorphic Computing and Engineering, Journal Year: 2024, Volume and Issue: 4(1), P. 014001 - 014001

Published: Jan. 10, 2024

Abstract Task specific hyperparameter tuning in reservoir computing is an open issue, and of particular relevance for hardware implemented reservoirs. We investigate the influence directly including externally controllable task timescales on performance sensitivity approaches. show that need optimisation can be reduced if are tailored to task. Our results mainly relevant temporal tasks requiring memory past inputs, example chaotic timeseries prediction. consider various methods approach demonstrate universality our message by looking at both time-multiplexed spatially-multiplexed computing.

Language: Английский

---

3D-integrated multilayered physical reservoir array for learning and forecasting time-series information

Nature Communications, Journal Year: 2024, Volume and Issue: 15(1)

Published: March 6, 2024

A wide reservoir computing system is an advanced architecture composed of multiple layers in parallel, which enables more complex and diverse internal dynamics for time-series information processing. However, its hardware implementation has not yet been realized due to the lack a high-performance physical complexity fabricating stacks. Here, we achieve proof-of-principle demonstration such made multilayered three-dimensional stacked 3 × 10 tungsten oxide memristive crossbar array, with further realize efficient learning forecasting data. Because three-layer structure allows seamless effective extraction intricate local features produced by various temporal inputs, it can readily outperform two-dimensional based approaches extensively studied previously. Our paves way systems capable efficiently processing dynamic information.

Language: Английский

---

Nanoscale magnonic networks

Physical Review Applied, Journal Year: 2024, Volume and Issue: 21(4)

Published: April 24, 2024

Spin waves and their quanta magnons are the collective excitations of a spin systems magnetic material, which offer potential for higher efficiency lower energy consumption in solving specific issues data processing. This Perspective discusses current challenges realizing magnonic circuits based on building blocks developed to date, further looks at application neuromorphic networks stochastic, reservoir, quantum computing, advantages over conventional electronics these areas.

Language: Английский

---

Nonlinearity in Memristors for Neuromorphic Dynamic Systems

Small Science, Journal Year: 2021, Volume and Issue: 2(1)

Published: Sept. 28, 2021

As semiconductor technology enters the more than Moore era, there exists an apparent contradiction between rapidly growing demands for data processing and visible inefficiency rooted in traditional computing architecture. Neuromorphic systems hold great prospects enabling a new generation of paradigm that can address this issue, which device components with rich dynamics nonlinearity. Herein, nonlinearity memristive devices their application building neuromorphic dynamic are reviewed. The internal mechanisms endow reviewed subsequently nonlinear spiking neurons implemented utilizing physical processes memristors shown. Typical examples on based summarized, including reservoir, oscillatory neural network, chaotic computing. Finally, outlook development is given.

Language: Английский

---