Neuromorphic Device With WOx/TiNOy Heterojunction for Airline Review Sentiment Prediction

Abstract

Neuromorphic device shows great potential in brain-like computing and is considered the hardware basis for breaking through the Von Neumann system of conventional computers. This article explores the application of WOx/TiNOy heterojunction neuromorphic devices in simulating biological synapse. The device successfully simulates a series of features of biological synapse, including paired-pulse facilitation (PPF), paired-pulse depression (PPD), long-term potential (LTP), long-time-range depression (LTD), continuous adjustable weights, and learning–forgetting–relearning based on the Schottky emission and Ag-conducting filament mechanisms. Using the device weights generated by the electrical impulses in conjunction with the long short-term memory (LSTM) network, the accuracy of the mood prediction task for the airline review dataset reached 99.23%, and the results demonstrated the potential of neuromorphic devices for application in the field of mood prediction.