Complex dynamics in chain HNN with parameter-relied equilibria and memristive electromagnetic induction

Area of Science:

• Computational Neuroscience
• Artificial Neural Networks
• Nonlinear Dynamics

Background:

• Understanding brain electrical activity requires investigating neural network dynamics.
• Electromagnetic induction plays a role in neuronal communication.
• Memristive devices offer novel approaches to neural network modeling.

Purpose of the Study:

• To propose and analyze a memristive chain Hopfield neural network (MCHNN) that incorporates electromagnetic induction.
• To explore the diverse dynamics and signal control properties of the MCHNN.
• To validate the MCHNN through hardware implementation and pseudorandomness testing.

Main Methods:

• Development of a memristive chain Hopfield neural network model with flux-controlled memristors.
• Numerical analysis of network equilibria and attractor dynamics (point, periodic, chaotic).
• Hardware platform construction for experimental verification and NIST statistical test suite application.

Main Results:

• The MCHNN exhibits diverse dynamical behaviors, including coexisting attractors, dependent on system parameters and initial conditions.
• The memristor's internal parameter effectively controls signal oscillation amplitude and flux properties.
• Hardware implementation validated the numerical findings, and NIST tests confirmed good pseudorandomness.

Conclusions:

• The proposed MCHNN successfully models electromagnetic induction between neurons.
• The MCHNN demonstrates rich dynamics and controllable signal properties, suitable for engineering applications.
• Experimental validation confirms the theoretical model's feasibility and potential for secure communication and random number generation.