Random network device fabricated using Ag₂Se nanowires for data augmentation with binarized convolutional neural network

Abstract An Ag 2 Se nanowire random network was fabricated for application as a data augmentation device and combined with binary convolutional neural (BCNN) to achieve high accuracy in voice classification tasks. Due the nonlinear high-dimensional characteristics resulting from formation of conductive filament at cross junction, could transform input into higher-order multiple signals, thereby enhancing task by augmenting signals. The results indicate that materials can realize same performance software, suggesting material-based hardware be used an elemental technology information processing.