Learning of an Artificial Neuron with Resistor-Memristor Synapses

The artificial neurons are important modules in the electronic devices and systems. Due to their widespread application, it is of high interest their new and efficient schematic realizations to be investigated. The purpose of this research is to suggest a comprehensive analysis of a modified memristor-based neuron with bridge memristor-resistor synapses. The analyzed in this paper device is based on a conventional neuron for noise suppression and resistor-memristor synapses. The applied memristor-based synaptic circuit is able to realize positive, zero and negative synaptic weights. For the computer simulations, a previously proposed by the authors in another research paper modified nonlinear drift memristor model is applied. Several main memristor models are also applied for the present investigation. A comparison between the results is made and a good matching between them is established. Advantages of the proposed synaptic circuit are the wide range of altering the synaptic weights, their simple tuning process by voltage pulses and the use of only two memristors and two nano-scale resistors.