Memristor-based circuit implementation of Competitive Neural Network based on online unsupervised Hebbian learning rule for pattern recognition

Li, Mian; Hong, Qinghui; Wang, Xiaoping

doi:10.1007/s00521-021-06361-4

Cited by 25 publications

(3 citation statements)

References 38 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This behavior has been previously associated with asymmetric Hebbian learning and has potential for applications in spiking neural network systems, where the frequency of a train of pulses is converted into a response of proportional intensity. 77 In both time-dependent measurements, PPF and STDP, the time intervals investigated for synaptic weight modulation are approximately 1 order of magnitude slower than biological time scales. 78 , 79 However, this limitation is introduced by the experimental setup rather than the actual devices, thus still showing promising use of Na + ion dynamics in biocompatible electronics.…”

Section: Results and Discussionmentioning

confidence: 99%

Sodium-Controlled Interfacial Resistive Switching in Thin Film Niobium Oxide for Neuromorphic Applications

Gaggio,

Jan,

Muller

et al. 2024

Chem. Mater.

View full text Add to dashboard Cite

A double layer 2-terminal device is employed to show Na-controlled interfacial resistive switching and neuromorphic behavior. The bilayer is based on interfacing biocompatible NaNbO3 and Nb2O5, which allows the reversible uptake of Na+ in the Nb2O5 layer. We demonstrate voltage-controlled interfacial barrier tuning via Na+ transfer, enabling conductivity modulation and spike-amplitude- and spike-timing-dependent plasticity. The neuromorphic behavior controlled by Na+ ion dynamics in biocompatible materials shows potential for future low-power sensing electronics and smart wearables with local processing.

show abstract

Section: Results and Discussionmentioning

confidence: 99%

Sodium-Controlled Interfacial Resistive Switching in Thin Film Niobium Oxide for Neuromorphic Applications

Gaggio,

Jan,

Muller

et al. 2024

Chem. Mater.

View full text Add to dashboard Cite

show abstract

“…Neural networks (NNs) plays an significant role in many real-world applications such as image encryption [1][2][3][4][5][6][7], signal processing [8], pattern recognition [9], optimization problem [10], and secure communication [11]. Cohen-Grossberg NNs (CGNNs) was initially proposed and studied by Cohen and Grossberg in 1983 [12].…”

Section: Introductionmentioning

confidence: 99%

Event-triggered controller on practically exponential input-to-state stabilization of stochastic reaction-diffusion neural networks and its application to image encryption

Gokulakrishnan

Sriniva

2023

Preprint

View full text Add to dashboard Cite

The stabilization problem for a class of stochastic reaction-diffusion delayed Cohen-Grossberg neural networks (SRDDCGNNs) with event-triggered controller is addressed in this paper. To address such a problem, Neumann boundary condition, distributed and boundary external disturbances are introduced. New sufficient criteria are derived by using the Lyapunov method, event-triggered mechanism, and the linear matrix inequality (LMI) approach to ensure the proposed controlled systems achieve practically exponential input-to-state stabilization. In light of these criteria, the impact of an event-triggered controller on practically exponential input-to-state stability (PEISS) is examined. Furthermore , the obtained results are successfully applied to stochastic reaction-diffusion delayed cellular neural networks (SRDDCNNs) and stochastic reaction-diffusion delayed Hopfield neural networks (SRD-DHNNs). At last, simulation results are given to illustrate the main results, and the SRDDHNNs are applied to the image encryption.

show abstract

“…The combination of neural networks with memristors [8], [9] exhibits remarkable properties such as fault tolerance [7] and event-triggered synchronization [10]. The design of such networks is an active area of research [11]- [13]. It is hence desirable to understand synchronization on a level relatively close to the hardware, for instance in terms of electrical parameters [14].…”

Section: Introductionmentioning

confidence: 99%

On the Electrical Modeling and Synchronization of Diffusively Coupled FitzHugh-Nagumo Oscillators

Lautenbacher¹,

Beattie²,

Ochs³

et al. 2022

Preprint

View full text Add to dashboard Cite

<p>We study the synchronization behavior of a class of identical FitzHugh-Nagumo-type oscillators under linear diffusive coupling. We describe the oscillators by a circuit model and we provide a sufficient synchronization condition that relies on the shape of the nonlinear conductance’s (i, u)-curve and the connectivity of the linear coupling network. We propose that this condition may have benefits with regard to the design of neuromorphic circuits using such oscillators and furthermore investigate how sharp our condition is. </p>

show abstract

Memristor-based circuit implementation of Competitive Neural Network based on online unsupervised Hebbian learning rule for pattern recognition

Cited by 25 publications

References 38 publications

Sodium-Controlled Interfacial Resistive Switching in Thin Film Niobium Oxide for Neuromorphic Applications

Sodium-Controlled Interfacial Resistive Switching in Thin Film Niobium Oxide for Neuromorphic Applications

Event-triggered controller on practically exponential input-to-state stabilization of stochastic reaction-diffusion neural networks and its application to image encryption

On the Electrical Modeling and Synchronization of Diffusively Coupled FitzHugh-Nagumo Oscillators

Contact Info

Product

Resources

About