ISCA Archive Interspeech 2023
ISCA Archive Interspeech 2023

EEG-based Auditory Attention Detection with Spatiotemporal Graph and Graph Convolutional Network

Ruicong Wang, Siqi Cai, Haizhou Li

The ability to detect auditory attention from electroencephalography (EEG) offers many possibilities for brain-computer interface (BCI) applications, such as hearing assistive devices. However, effective feature representation for EEG signals remains a challenge due to the complex spatial and temporal dynamics of EEG signals. To overcome this challenge, we introduce a Spatiotemporal Graph Convolutional Network (ST-GCN), which combines a temporal attention mechanism and a graph convolutional module. The temporal attention mechanism captures the temporal dynamics of EEG segments, while the graph convolutional module learns the spatial pattern of multi-channel EEG signals. We evaluate the performance of our proposed ST-GCN on two publicly available datasets and demonstrate significant improvements over existing state-of-the-art models. These findings suggest that the ST-GCN model has the potential to advance auditory attention detection in real-life BCI applications.