Bioinspired Memristive Neural Network Circuit Design of Cross-Modal Associative Memory

Abstract

The development of brain-like artificial intelligence is based on the cognitive functions of the brain, which are influenced by the cross-modal interactions of learning and memory. Inspired by the neural mechanism and biological phenomena of associative memory in Drosophila, this article proposes a bioinspired neural network and memristive circuit of cross-modal associative memory to mimic the cross-modal interaction during brain association. The designed circuit mainly consists of three modules: 1) the threshold module, which uses to determine the threshold level of conditioned stimuli (CS) that the brain can use to trigger learning and memory effects; 2) the synergy module, which performs cross-modal synergy processing of CS signals; and 3) the synapse and neuron module, which uses memristors to mimic synaptic weights and output neurons for associative learning. According to the different intensities of the input stimuli signals under conditioning, the proposed circuit implements the innovative functions, such as unimodal learning with threshold, cross-modal reinforcement, cross-modal facilitation, and cross-modal memory transfer. The simulation results in PSPICE show that the proposed circuit exhibits cross-modal synergy and transfer interactions, and it also provides further references for the research and development of bionic intelligent robots and brain-like intelligence.