One-Board Design and Simulation of Double-Layer Perceptron Based on Metal-Oxide Memristive Nanostructures

Design and training principles have been proposed and tested for an artificial neural network based on metal-oxide thin-film nanostructures possessing bipolar resistive switching (memristive) effect. Experimental electronic circuit of neural network is implemented as a double-layer perception with a weight matrix composed of 32 memristive devices. The network training algorithm takes into account technological variations of the parameters of memristive devices. Despite the limited size of weight matrix the developed neural network model is scalable and capable of solving nonlinear classification problems. The learning and functionality of the network are demonstrated by using its computer model for the classification of activity propagation directions in simulated neuronal culture.