Multi-mechanism driven geometric control of discrete memristive dual-neuron HNN: Modulation analysis and hardware...

Area of Science:

• Neuroscience
• Complex Systems
• Computational Neuroscience

Background:

• Dynamical modulation mechanisms in discrete memristive Hopfield neural networks (HNNs) are a key research area.
• Understanding synaptic crosstalk effects is crucial for advancing neural network dynamics.

Purpose of the Study:

• To propose and investigate a four-dimensional discrete Hopfield neural network model (4DMCHNN) incorporating synaptic crosstalk.
• To analyze the complex dynamical regulatory behaviors and chaotic properties influenced by synaptic crosstalk.

Main Methods:

• Development of a four-dimensional discrete Hopfield neural network model (4DMCHNN).
• Systematic investigation of dynamical regulatory behaviors under synaptic crosstalk.
• Numerical simulations to analyze chaotic phenomena, including amplitude control and periodic modulation.
• Implementation on a digital circuit platform and validation using NIST statistical tests for a pseudo-random number generator.

Main Results:

• The 4DMCHNN exhibits rich chaotic phenomena, with amplitude control dependent on synaptic crosstalk intensity and memristor parameters.
• Periodic dynamic modulation is primarily influenced by memristor parameters, enhancing attractor offset regulation.
• The system demonstrates initial-value-induced shifts and coexistence of homogeneous attractors.
• A pseudo-random number generator based on the 4DMCHNN passed NIST statistical tests, indicating hardware applicability.

Conclusions:

• Synaptic crosstalk significantly influences the chaotic dynamics of discrete memristive Hopfield neural networks.
• The proposed 4DMCHNN offers predictable modulation of chaotic behaviors, suitable for neuromorphic applications and chaos control.
• Hardware implementation of the 4DMCHNN demonstrates its potential for practical, low-cost applications.