Synchronously adjustable offset boosting and amplitude control in memristive neural network with hardware implementation

Traditional dynamic systems can exhibit either amplitude control or offset boosting, but not both simultaneously, thereby limiting their applicability in specific engineering applications. This paper introduces a novel ReLU-type local active memristor (LAM)-based Hopfield neural network (HNN) model and reveals that synchronously adjustable offset boosting and amplitude control occur via system parameters, which have not been presented in neural network systems, yet. Numerical analyses confirm the emergence of chaos, bifurcation, and attractor control, and demonstrate that memristor coupling strength, internal parameters, and increment parameters all can simultaneously achieve both offset boosting and amplitude control. Analog circuit simulations and digital hardware experiments are developed, validating the practical applicability of the adjustable control. Finally, the application of image encryption based on DNA coding is realized, and the results show that the designed algorithm is effective. This work provides a new approach for regulating neural network models using memristors.