Abstract:Memristors are of great theoretical and practical significance for chaotic dynamics research of brain-like neural networks due to their excellent physical properties such as brain synapse-like memorability and nonlinearity, especially crucial for the promotion of AI big models, cloud computing, and intelligent systems in the artificial intelligence field. In this paper, we introduce memristors as self-connecting synapses into a four-dimensional Hopfield neural network, constructing a central cyclic memristive neural network (CCMNN), and achieving its effective control. The model adopts a central loop topology and exhibits a variety of complex dynamic behaviors such as chaos, bifurcation, and homogeneous and heterogeneous coexisting attractors. The complex dynamic behaviors of the CCMNN are investigated in depth numerically by equilibrium point stability analysis as well as phase trajectory maps, bifurcation maps, time-domain maps, and LEs. It is found that with the variation of the internal parameters of the memristor, asymmetric heterogeneous attractor coexistence phenomena appear under different initial conditions, including the multi-stable coexistence behaviors of periodic-periodic, periodic-stable point, periodic-chaotic, and stable point-chaotic. In addition, by adjusting the structural parameters, a wide range of amplitude control can be realized without changing the chaotic state of the system. Finally, based on the CCMNN model, an adaptive synchronization controller is designed to achieve finite-time synchronization control, and its application prospect in simple secure communication is discussed. A microcontroller-based hardware circuit and NIST test are conducted to verify the correctness of the numerical results and theoretical analysis.

Winner-Takes-All Mechanism Realized By Memristive Neural Network

Analysis and Design of Winner-Take-All Behavior Based on A Novel Memristive Neural Network

Competitive Neural Network Circuit Based on Winner-Take-All Mechanism and Online Hebbian Learning Rule.

A Winner-Take-All Neural Networks of N Linear Threshold Neurons Without Self-Excitatory Connections

Selective positive-negative feedback produces the winner-take-all competition in recurrent neural networks.

Dual functional states of working memory realized by memristor-based neural network

Versatile networks of simulated spiking neurons displaying winner-take-all behavior

Neuromorphic Implementation of Attractor Dynamics in a Two-Variable Winner-Take-All Circuit with NMDARs: A Simulation Study

A Novel Recurrent Neural Network With One Neuron and Finite-Time Convergence for -Winners-Take-All Operation

Winner-Take-All as Basic Probabilistic Inference Unit of Neuronal Circuits

A neural network circuit with associative learning and forgetting process based on memristor neuromorphic device

A Model of Neural Circuit with Adaption and Nonlinear Connection

Region stability analysis and tracking control of memristive recurrent neural network

Constructing Memristive Hindmarsh–Rose Neuron with Countless Coexisting Firings

Neuromorphic Dynamics Near the Edge of Chaos in Memristive Neurons

A Memristor-Based Circuit with the Loser-Take-All Mechanism for Classification

A novel memristive synapse-coupled ring neural network with countless attractors and its application

Memristor-Based Neuromorphic System With Content Addressable Memory Structure

Realization of the conscience mechanism in CMOS implementation of winner-takes-all self-organizing neural networks.

Memristive Single-Neuron Model and Its Memristor-Coupled Network: Homogenously Coexisting Attractors and Parallel-Offset Synchronization

Heterogeneous coexisting attractors, large-scale amplitude control and finite-time synchronization of central cyclic memristive neural networks