Synchronization of Analog Neuron Circuits With Digital Memristive Synapses: An Hybrid Approach

Abstract

The realization of hybrid (analog/digital) circuits mimicking the nature of interconnected neural units represents a step toward control engineering practical applications of neural networks. In fact, while analog neurons provide complete flexibility and ensure robustness to uncertainty and noise, the implementation of a digital coupling interface guarantees the full reconfigurability of interconnection networks. The hybrid implementation, therefore, ensures control actions reliable in practical scenarios, ranging from robotics to process control. In this paper, the synchronized behavior of a pair of analog circuits designed from the Izhikevich neuron model, coupled through a digitally implemented memristive synapse, is discussed from numerical and experimental perspectives. The results pave the way for the implementation of self-organizing and adaptive control strategies.