Can memristors revolutionize neuromorphic computing? This paper explores the potential of Nb2O5 memristors to construct artificial neurons in unsupervised learning networks, offering a pathway to nanoscale and low-power computing solutions. They find that memristors can be used to construct artificial neurons. Researchers apply an Nb2O5 locally active memristor with parasitic capacitance as a LIF neuron, analyzing spiking oscillations through small signal equivalent circuits and the Hopf bifurcation method. By combining Nb2O5 memristive neurons with voltage-controlled non-volatile memristive synapses, they create an unsupervised learning network. The proposed circuit, validated through LTspice simulation, demonstrates the ability to recognize different patterns and can be applied to the neural morphological system of pattern recognition and classify letter and number images. Before building the hardware circuit, we predict the training time, recognition time, and recognition accuracy of the pattern recognition network through theoretical analysis, which guides the actual circuit experiment.
This article, published in the International Journal of Circuit Theory and Applications, aligns with the journal’s focus on electronics and electrical engineering. By exploring the application of Nb2O5 memristors in neuromorphic computing, the research contributes to advancements in circuit theory and the development of innovative electronic devices. The journal’s scope.