Student or educator? Sign up with your school email and get 6 months of Pro free.
CoChat

The DOI matches its Crossref record; 51 of the 51 references it cites were found in a registry or an index.

Spiking neurons with spatiotemporal dynamics and gain modulation for monolithically integrated memristive neural networks

Open access
Duan, Qingxi, Jing, Zhaokun, Zou, Xiaolong, Wang, Yanghao, Yang, Ke, Zhang, Teng, Wu, Si, Huang, Ru, Yang, Yuchao · 2020 · Nature Communications

Abstract

As a key building block of biological cortex, neurons are powerful information processing units and can achieve highly complex nonlinear computations even in individual cells. Hardware implementation of artificial neurons with similar capability is of great significance for the construction of intelligent, neuromorphic systems. Here, we demonstrate an artificial neuron based on NbO x volatile memristor that not only realizes traditional all-or-nothing, threshold-driven spiking and spatiotemporal integration, but also enables dynamic logic including XOR function that is not linearly separable and multiplicative gain modulation among different dendritic inputs, therefore surpassing neuronal functions described by a simple point neuron model. A monolithically integrated 4 × 4 fully memristive neural network consisting of volatile NbO x memristor based neurons and nonvolatile TaO x memristor based synapses in a single crossbar array is experimentally demonstrated, showing capability in pattern recognition through online learning using a simplified δ-rule and coincidence detection, which paves the way for bio-inspired intelligent systems.

Overview

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1038/s41467-020-17215-3

Cites

51 of 51 references verified0 citation errors · 0 not checkable

Statistics

Figures

Fig. 1 Comparison of biological and artificial neurons. Schematic of biological neurons and synapses (left) compared with an artificial neural network (right).
Fig. 1 Comparison of biological and artificial neurons. Schematic of biological neurons and synapses (left) compared with an artificial neural network (right).CC BY · PubMed Central
Fig. 2 Spiking neuron based on NbOx volatile memristors. a Schematic diagram of the memristive device, which consists of a NbOx layer sandwiched between two Pt/Ti electrodes. b Current-voltage characteristics of the device repeated for 10 cycles. c Schematic illustration of the mechanism responsible for threshold switching in the NbOx devices. d Illustration of an ion channel embedded in cell membrane of a biological neuron, with corresponding circuit based on NbOx devices for implementation of a spiking neuron. e Characterization of the LIF neuron under a continuous pulse train and the influence of varying capacitance (Cm) and resistance (RL). f The firing response of the spiking neuron under different input and circuit conditions. When changing the input parameters such as the pulse width, pulse interval and circuit parameters such as Cm and RL, the firing pulse count has an obvious change with 20 input pulse cycles.
Fig. 2 Spiking neuron based on NbOx volatile memristors. a Schematic diagram of the memristive device, which consists of a NbOx layer sandwiched between two Pt/Ti electrodes. b Current-voltage characteristics of the device repeated for 10 cycles. c Schematic illustration of the mechanism responsible for threshold switching in the NbOx devices. d Illustration of an ion channel embedded in cell membrane of a biological neuron, with corresponding circuit based on NbOx devices for implementation of a spiking neuron. e Characterization of the LIF neuron under a continuous pulse train and the influence of varying capacitance (Cm) and resistance (RL). f The firing response of the spiking neuron under different input and circuit conditions. When changing the input parameters such as the pulse width, pulse interval and circuit parameters such as Cm and RL, the firing pulse count has an obvious change with 20 input pulse cycles.CC BY · PubMed Central
Fig. 3 Spatiotemporal integration and dynamic logic in NbOx based neuron. a The circuit diagram of the spiking neuron. b Schematic diagram of spatial summation with different input pulse amplitudes. c Neuronal response triggered by two input pulses (0.9 V amplitude, 1 μs width, interval 0.1 μs, 10 pulse cycles) applied on S1 and S2 individually and simultaneously. d The spatial summation at varied conditions and corresponding fitting results. The amplitudes of the two input spikes (V1 and V2) were systematically changed from 0.6 to 0.9 V, and applied simultaneously to the respective nodes. e The input–output characteristics for “AND” logic with two input pulses applied on S1 and S2. f Schematic diagram of spatial summation with different input pulse intervals. g Neuronal response triggered by two input pulses (0.8 V amplitude, 1 μs width, interval 0.1 μs, 10 pulse cycles) applied on S1 and S2 individually and simultaneously. h The spatial summation at varied conditions and corresponding fitting results. The intervals of the two input spikes were systematically changed from 0.05 to 0.4 μs. i The input–output characteristics for “XOR” logic with two input pulses applied on S1 and S2. j Schematic diagram of spatiotemporal summation with different time intervals of input pulses. k Neuronal response triggered by two input pulses (0.9 V amplitude, 1 μs width, 0.1 μs interval) applied on S1 and S2 with different time intervals of 0, 2.2 and −2.2 μs. l Spatiotemporal summation results when the time interval of two input spikes is changed from −6.6 to 6.6 μs. The solid line is fitting result by Eq. (1).
Fig. 3 Spatiotemporal integration and dynamic logic in NbOx based neuron. a The circuit diagram of the spiking neuron. b Schematic diagram of spatial summation with different input pulse amplitudes. c Neuronal response triggered by two input pulses (0.9 V amplitude, 1 μs width, interval 0.1 μs, 10 pulse cycles) applied on S1 and S2 individually and simultaneously. d The spatial summation at varied conditions and corresponding fitting results. The amplitudes of the two input spikes (V1 and V2) were systematically changed from 0.6 to 0.9 V, and applied simultaneously to the respective nodes. e The input–output characteristics for “AND” logic with two input pulses applied on S1 and S2. f Schematic diagram of spatial summation with different input pulse intervals. g Neuronal response triggered by two input pulses (0.8 V amplitude, 1 μs width, interval 0.1 μs, 10 pulse cycles) applied on S1 and S2 individually and simultaneously. h The spatial summation at varied conditions and corresponding fitting results. The intervals of the two input spikes were systematically changed from 0.05 to 0.4 μs. i The input–output characteristics for “XOR” logic with two input pulses applied on S1 and S2. j Schematic diagram of spatiotemporal summation with different time intervals of input pulses. k Neuronal response triggered by two input pulses (0.9 V amplitude, 1 μs width, 0.1 μs interval) applied on S1 and S2 with different time intervals of 0, 2.2 and −2.2 μs. l Spatiotemporal summation results when the time interval of two input spikes is changed from −6.6 to 6.6 μs. The solid line is fitting result by Eq. (1).CC BY · PubMed Central
Fig. 4 Neuronal gain modulation of NbOx based neuron. a Schematic diagram of the rate-coded neuronal signaling. The driving input and modulatory input affect output firing rate jointly. b Schematic diagram of neuronal gain modulation by modulatory inputs, including input modulation (orange line) and output modulation (purple line). The blue line is the I–O curve with only the driving input. c Neuronal responses triggered by only the driving input (1.2 V amplitude, 1 μs width, 0.4 μs interval) applied on S1 (bottom), and triggered by the driving input (1.2 V amplitude) and modulatory input (1 μs width, 0.4 μs interval) with different amplitudes of 0.4 V, 0.5 V, 0.6 V and 0.7 V, applied simultaneously to nodes S1 and S2, as shown in the middle and top of the panel. d Experimental neuron responses showing multiplicative gain modulation when applying different modulatory inputs (m = 0, 0.4, 0.5, 0.6 and 0.7 V). The background lines indicate the overall trend in data. e Fitting lines calculated by Eq. (2).
Fig. 4 Neuronal gain modulation of NbOx based neuron. a Schematic diagram of the rate-coded neuronal signaling. The driving input and modulatory input affect output firing rate jointly. b Schematic diagram of neuronal gain modulation by modulatory inputs, including input modulation (orange line) and output modulation (purple line). The blue line is the I–O curve with only the driving input. c Neuronal responses triggered by only the driving input (1.2 V amplitude, 1 μs width, 0.4 μs interval) applied on S1 (bottom), and triggered by the driving input (1.2 V amplitude) and modulatory input (1 μs width, 0.4 μs interval) with different amplitudes of 0.4 V, 0.5 V, 0.6 V and 0.7 V, applied simultaneously to nodes S1 and S2, as shown in the middle and top of the panel. d Experimental neuron responses showing multiplicative gain modulation when applying different modulatory inputs (m = 0, 0.4, 0.5, 0.6 and 0.7 V). The background lines indicate the overall trend in data. e Fitting lines calculated by Eq. (2).CC BY · PubMed Central
Fig. 5 Fully memristive spiking neural network. a Scanning electron microscopy (SEM) image of the monolithically integrated memristive neural network. Scale bar, 100 μm. b SEM image of the Pt/Ti/NbOx/Pt threshold switching device. Scale bar, 20 μm. c Cross-sectional transmission electron microscopy (TEM) image of the Pt/Ti/NbOx/Pt threshold switching device. Scale bar, 50 nm. d SEM image of a Pt/Ta/TaOx/Pt synapse device. Scale bar, 10 μm. e Cross-sectional TEM image of the Pt/Ta/TaOx/Pt device. Scale bar, 20 nm. f, g Neuronal outputs when presented with different input patterns. f Neuron 1 learnt to recognize pattern “0101”, while neuron 2 learnt to recognize pattern “1100” g. The input pulses are 1 V in amplitude, 1 μs in width, 0.1 μs in interval and repeated for 10 cycles. h Schematic illustration of supervised learning in a fully memristive neural network. i Evolutions of synaptic weights over time during online learning, where the pattern “1010” is repeatedly applied, with an initial state of “0110”.
Fig. 5 Fully memristive spiking neural network. a Scanning electron microscopy (SEM) image of the monolithically integrated memristive neural network. Scale bar, 100 μm. b SEM image of the Pt/Ti/NbOx/Pt threshold switching device. Scale bar, 20 μm. c Cross-sectional transmission electron microscopy (TEM) image of the Pt/Ti/NbOx/Pt threshold switching device. Scale bar, 50 nm. d SEM image of a Pt/Ta/TaOx/Pt synapse device. Scale bar, 10 μm. e Cross-sectional TEM image of the Pt/Ta/TaOx/Pt device. Scale bar, 20 nm. f, g Neuronal outputs when presented with different input patterns. f Neuron 1 learnt to recognize pattern “0101”, while neuron 2 learnt to recognize pattern “1100” g. The input pulses are 1 V in amplitude, 1 μs in width, 0.1 μs in interval and repeated for 10 cycles. h Schematic illustration of supervised learning in a fully memristive neural network. i Evolutions of synaptic weights over time during online learning, where the pattern “1010” is repeatedly applied, with an initial state of “0110”.CC BY · PubMed Central
Fig. 6 A 3-layer spiking neural network by simulation. a Schematic of the spiking neural network for MNIST classification. Input images are first converted to Poisson spike trains, where the spiking rates are proportional to pixel values. The spike trains are then input to the network, weighted and integrated by the neurons. At last, we count the spiking rates of output neurons to get the prediction results. b Flow chart of the simulation process. In the forward phase (yellow arrow), input spike trains are weighted by the memristor crossbar and then integrated on neurons, and the spiking rates of output neurons are used in loss computing. In the backward phase (purple arrow), a soft function like sigmoid is used as an alternative to neuron spike function (step function) in gradient computing. The computing units in green boxes are simulated based on experimental data, while the units in blue boxes are implemented by software. c In the case where the input picture is correctly identified, the statistics of the firing numbers issued by ten neurons in the case of input pictures corresponding to the MNIST numbers themselves. Statistics of firing numbers of the category neurons in response to different inputs, showing that the inputs are classified correctly in most cases. d Averaged confusion matrix of the testing results, showing that the test inputs are well classified after training.
Fig. 6 A 3-layer spiking neural network by simulation. a Schematic of the spiking neural network for MNIST classification. Input images are first converted to Poisson spike trains, where the spiking rates are proportional to pixel values. The spike trains are then input to the network, weighted and integrated by the neurons. At last, we count the spiking rates of output neurons to get the prediction results. b Flow chart of the simulation process. In the forward phase (yellow arrow), input spike trains are weighted by the memristor crossbar and then integrated on neurons, and the spiking rates of output neurons are used in loss computing. In the backward phase (purple arrow), a soft function like sigmoid is used as an alternative to neuron spike function (step function) in gradient computing. The computing units in green boxes are simulated based on experimental data, while the units in blue boxes are implemented by software. c In the case where the input picture is correctly identified, the statistics of the firing numbers issued by ten neurons in the case of input pictures corresponding to the MNIST numbers themselves. Statistics of firing numbers of the category neurons in response to different inputs, showing that the inputs are classified correctly in most cases. d Averaged confusion matrix of the testing results, showing that the test inputs are well classified after training.CC BY · PubMed Central
Fig. 7 Coincidence detection based on spatiotemporal dynamics. a Schematic diagram of coincidence detection in the fully memristive neural network. b The neuronal response by two synchronous input pulse trains (1.1 V in amplitude, 1 μs in width, 1.5 μs in interval, for 50 cycles) applied on S1 and S2. c The neuronal response by two asynchronous input pulse trains (1.1 V in amplitude, 1 μs in width, 1.5 μs in interval, for 50 cycles) applied on S1 and S2, where S2 is behind S1 by 1.2 μs. d The neuronal response by two asynchronous input pulse trains (1.1 V in amplitude, 1 μs in width, for 50 cycles) applied on S1 and S2, where S1 has and interval of 1.5 μs and S2 is random relative to S2 in timing. e Input pulse trains of the neuron from independent 4000 excitatory and 1000 inhibitory random spike trains following Poisson statistics. f Introduction of synchronous events following Poisson statistics into the excitatory inputs, where the input rates are unchanged and the proportion of synchronous events is 0.3%. g Simulated neuronal response as a result of the inputs shown in e, where the artificial neuron only fires 2 spikes. h Simulated neuronal response as a result of the inputs shown in f, where the artificial neuron fires 21 spikes.
Fig. 7 Coincidence detection based on spatiotemporal dynamics. a Schematic diagram of coincidence detection in the fully memristive neural network. b The neuronal response by two synchronous input pulse trains (1.1 V in amplitude, 1 μs in width, 1.5 μs in interval, for 50 cycles) applied on S1 and S2. c The neuronal response by two asynchronous input pulse trains (1.1 V in amplitude, 1 μs in width, 1.5 μs in interval, for 50 cycles) applied on S1 and S2, where S2 is behind S1 by 1.2 μs. d The neuronal response by two asynchronous input pulse trains (1.1 V in amplitude, 1 μs in width, for 50 cycles) applied on S1 and S2, where S1 has and interval of 1.5 μs and S2 is random relative to S2 in timing. e Input pulse trains of the neuron from independent 4000 excitatory and 1000 inhibitory random spike trains following Poisson statistics. f Introduction of synchronous events following Poisson statistics into the excitatory inputs, where the input rates are unchanged and the proportion of synchronous events is 0.3%. g Simulated neuronal response as a result of the inputs shown in e, where the artificial neuron only fires 2 spikes. h Simulated neuronal response as a result of the inputs shown in f, where the artificial neuron fires 21 spikes.CC BY · PubMed Central
Fig. 8 A network of memristive neurons with gain modulation for receptive field remapping. a A one-dimensional network where neurons are connected in an uni-direction. Each neuron receives a modulatory signal (orange arrow). A visual input is applied at the future receptive field (FRF) of a neuron (the yellow one). b The visual input at FRF and the modulatory signal (top, red line) triggers a propagation of neural activity from FRF to current receptive field of the yellow neuron, as if the receptive field of the yellow neuron is temporally expanded when the modulatory signal is applied, achieving RF remapping (right panel). The distance of remapping is controlled by the duration of the modulatory signal. In the simulation, the visual input is applied during 0−300 ms, while the modulatory signal m = 0.8 is applied during 300−1000 ms. θthr = 0.41, Wi,i-1 = 0.095, τ = 10 ms, dt = 1 ms, Iext,0 = 0.09.
Fig. 8 A network of memristive neurons with gain modulation for receptive field remapping. a A one-dimensional network where neurons are connected in an uni-direction. Each neuron receives a modulatory signal (orange arrow). A visual input is applied at the future receptive field (FRF) of a neuron (the yellow one). b The visual input at FRF and the modulatory signal (top, red line) triggers a propagation of neural activity from FRF to current receptive field of the yellow neuron, as if the receptive field of the yellow neuron is temporally expanded when the modulatory signal is applied, achieving RF remapping (right panel). The distance of remapping is controlled by the duration of the modulatory signal. In the simulation, the visual input is applied during 0−300 ms, while the modulatory signal m = 0.8 is applied during 300−1000 ms. θthr = 0.41, Wi,i-1 = 0.095, τ = 10 ms, dt = 1 ms, Iext,0 = 0.09.CC BY · PubMed Central

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

Show all 49 references

Research works printed but not linked: 1

  1. Minsky, M. L. & Papert, S. A. Perceptrons. (MIT Press, Cambridge, MA, USA, 1969).

Cited by

See all 339

Metadata

Authors
Duan, Q., Jing, Z., Zou, X., Wang, Y., Yang, K., Zhang, T., Wu, S., Huang, R., & Yang, Y.
Affiliations
Peking University (CN)
Corresponding author
  • Huang, Ru
  • Yang, Yuchao
Venue
Nature Communications
Published
2020-07-07
Volume
11
Issue
1
Pages
3399-3399
ISSN
2041-1723 (ISSN-L)
OpenAlex
W3039560462
Cited by
339
References
49
Citations by year
2026: 35 · 2025: 63 · 2024: 71 · 2023: 62 · 2022: 71 · 2021: 37
FWCI
14.60
Type
journal article
Language
en
Open access
Licence
Locations
  • Nature Communications (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • RePEc: Research Papers in Economics (submittedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
  • Greater South Information System
Funders
  • National Natural Science Foundation of China (61927901, 61421005, 61674006)
  • National Outstanding Youth Foundation of China (61925401)
Topics
  • Physical Sciences › Engineering › Electrical and Electronic Engineering › Advanced Memory and Neural Computing
  • Physical Sciences › Engineering › Electrical and Electronic Engineering › Ferroelectric and Negative Capacitance Devices
  • Physical Sciences › Computer Science › Artificial Intelligence › Neural Networks and Applications
Keywords
spiking neurons, memristive neural networks, spatiotemporal integration, gain modulation, XOR function, monolithic integration, crossbar arrays, coincidence detection, pattern recognition, online learning, neuromorphic hardware, artificial neurons

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: Spiking neurons with spatiotemporal dynamics and gain modulation for monolithically integrated memristive neural networks. https://app.cochat.ai/papers/10.1038/s41467-020-17215-3
The paper
Duan, Q., Jing, Z., Zou, X., Wang, Y., Yang, K., Zhang, T., Wu, S., Huang, R., & Yang, Y. (2020). Spiking neurons with spatiotemporal dynamics and gain modulation for monolithically integrated memristive neural networks. Nature Communications. https://doi.org/10.1038/s41467-020-17215-3

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.