Reservoir computing using networks of memristors: effects of topology and heterogeneity

Mallinson, Josh; Heywood, Zachary E.; Daniels, Ryan. K.; Arnold, Matthew D.; Bones, Philip J.; Brown, S. A.

doi:10.1039/d2nr07275k

Cited by 7 publications

(21 citation statements)

References 48 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…[ 110–113 ] The network gap formed by such random deposition can be regarded as random network with changing resistive distribution. [ 114,115 ] Changes in voltage pulses induce the formation and annihilation of interstitial atomic lines in the network, and multiple switching events at different locations lead to complex switching dynamics throughout the network.…”

Section: Physical Reservoir Computing By Nanoscale Materials and Devicesmentioning

confidence: 99%

Physical Reservoir Computing Based on Nanoscale Materials and Devices

Qi,

Mi,

Qian

et al. 2023

Adv Funct Materials

View full text Add to dashboard Cite

Bioinspired computation systems can achieve artificial intelligence, bypassing fundamental bottlenecks and cost constraints. Computational frameworks suited for temporal/sequential data processing such as recurrent neural networks (RNNs) suffer from problems of high complexity and low efficiency. Physical systems assembled with nanoscale materials and devices represent as an alternative route to serve as the core component for physically implanted reservoir computing. In this review, an overview of the development of the paradigm of physical reservoir computing (PRC) is provided and the typical physical reservoirs constructed with nanomaterials and nanodevices are described. The physical reservoirs based on multiple nanomaterials overcome the problems of RNN, show strong robustness, and effectively deal with tasks with improved reliability and availability. Finally, the challenges and perspectives of nanomaterial and nanodevice‐based PRC as a component of next‐generation machine learning systems are discussed.

show abstract

Section: Physical Reservoir Computing By Nanoscale Materials and Devicesmentioning

confidence: 99%

Physical Reservoir Computing Based on Nanoscale Materials and Devices

Qi,

Mi,

Qian

et al. 2023

Adv Funct Materials

View full text Add to dashboard Cite

show abstract

“…Nanoparticle (NP) networks as one example of physical systems (Dale et al, 2021;Milano et al, 2023) offer a promising avenue in this field. One current approach uses percolating scale-free NP networks with small-world properties (Fostner and Brown, 2015;Bose et al, 2017;Daniels et al, 2023;Mallinson et al, 2023). The intrinsic architecture of these networks includes conductivity dynamics which are crucial for computational tasks (Deng and Zhang, 2007;Kawai et al, 2019).…”

Section: Introductionmentioning

confidence: 99%

A kinetic Monte Carlo approach for Boolean logic functionality in gold nanoparticle networks

Mensing,

van der Wiel,

Heuer

2024

Front. Nanotechnol.

View full text Add to dashboard Cite

Nanoparticles interconnected by insulating organic molecules exhibit nonlinear switching behavior at low temperatures. By assembling these switches into a network and manipulating charge transport dynamics through surrounding electrodes, the network can be reconfigurably functionalized to act as any Boolean logic gate. This work introduces a kinetic Monte Carlo-based simulation tool, applying established principles of single electronics to model charge transport dynamics in nanoparticle networks. We functionalize nanoparticle networks as Boolean logic gates and assess their quality using a fitness function. Based on the definition of fitness, we derive new metrics to quantify essential nonlinear properties of the network, including negative differential resistance and nonlinear separability. These nonlinear properties are crucial not only for functionalizing the network as Boolean logic gates but also when our networks are functionalized for brain-inspired computing applications in the future. We address fundamental questions about the dependence of fitness and nonlinear properties on system size, number of surrounding electrodes, and electrode positioning. We assert the overall benefit of having more electrodes, with proximity to the network’s output being pivotal for functionality and nonlinearity. Additionally, we demonstrate an optimal system size and argue for breaking symmetry in electrode positioning to favor nonlinear properties.

show abstract

“…Self-organized nanoscale networks have recently been demonstrated to have a variety of brain-like characteristics. These networks (comprising, e.g., nanowires, [1][2][3][4] nanotubes, [5][6][7] or DOI: 10.1002/adma.202402319 nanoparticles [8][9][10][11][12] ) combine large numbers of functional elements (synapses and/or neurons) into brain-like architectures and are fabricated with simple and cost-effective processes. They have the potential to perform complex computational tasks with low power consumption, and are therefore appealing systems for neuromorphic computation.…”

Section: Introductionmentioning

confidence: 99%

Experimental Demonstration of Reservoir Computing with Self‐Assembled Percolating Networks of Nanoparticles

Mallinson,

Steel,

Heywood

et al. 2024

Advanced Materials

Self Cite

View full text Add to dashboard Cite

The complex self‐assembled network of neurons and synapses that comprizes the biological brain enables natural information processing with remarkable efficiency. Percolating Networks of Nanoparticles (PNNs) are complex self‐assembled nanoscale systems that have been shown to possess many promising brain‐like attributes and which are therefore appealing systems for neuromorphic computation. Here we perform experiments that show that PNNs can be utilized as physical reservoirs within a nanoelectronic reservoir computing framework and demonstrate successful computation for several benchmark tasks (chaotic time series prediction, non‐linear transformation and memory capacity). For each task we compile relevant literature results and show that the performance of the PNNs compares favourably to that previously reported from nanoelectronic reservoirs. We then demonstrate experimentally that PNNs can be used for spoken digit recognition with state‐of‐the‐art accuracy. Finally, we emulate a parallel reservoir architecture, which increases the dimensionality and richness of the reservoir outputs and results in further improvements in performance across all tasks.This article is protected by copyright. All rights reserved

show abstract

Reservoir computing using networks of memristors: effects of topology and heterogeneity

Abstract: Reservoir computing (RC) has attracted significant interest as a framework for the implementation of novel neuromorphic computing architectures. Previously attention has been focussed on software-based reservoirs, where it has been...

Cited by 7 publications

References 48 publications

Physical Reservoir Computing Based on Nanoscale Materials and Devices

Physical Reservoir Computing Based on Nanoscale Materials and Devices

A kinetic Monte Carlo approach for Boolean logic functionality in gold nanoparticle networks

Experimental Demonstration of Reservoir Computing with Self‐Assembled Percolating Networks of Nanoparticles

Contact Info

Product

Resources

About