A Neural-Inspired Architecture for EEG-Based Auditory Attention Detection

Humans have the ability to focus on one of sound sources in a noisy scene, which is critical for everyday communication. Auditory attention detection (AAD) seeks detect selective from one’s brain signals. For AAD be useful brain–computer interface applications, new approaches with low computational cost, high classification performance, and latency are required developed. In this study, we proposed novel neural-inspired architecture mimic neural computation coding strategy electroencephalography-based AAD. We validated our model through data visualization, conducted experiments two publicly available databases. both KUL DTU databases, it outperforms linear convolutional network (CNN) models consistent improvements 1 s 5 decision windows terms accuracy. Although accuracy inferior state-of-the-art spatio-spectral feature (SSF)-CNN model, cost less than 1% SSF-CNN’s. Moreover, decoder more hardware friendly energy-efficient due its biological computing scheme. Overall, realizes fast, accurate, energy expenditure AAD, big step forward towards practical neuro-steered hearing aids.