A Multi-Plane Decoupled Convolutional Network for EEG-Based Auditory Attention Detection
Paper
Huang, Wei, Mei, Jiahao, Wei, Shicheng and Wang, Huabin. 2024. "A Multi-Plane Decoupled Convolutional Network for EEG-Based Auditory Attention Detection." 2024 9th International Conference on Signal and Image Processing (ICSIP). Nanjing, China 12 - 14 Jul 2024 China. IEEE (Institute of Electrical and Electronics Engineers). https://doi.org/10.1109/ICSIP61881.2024.10671502
| Paper/Presentation Title | A Multi-Plane Decoupled Convolutional Network for EEG-Based Auditory Attention Detection |
|---|---|
| Presentation Type | Paper |
| Authors | Huang, Wei, Mei, Jiahao, Wei, Shicheng and Wang, Huabin |
| Journal or Proceedings Title | Proceedings of 2024 9th International Conference on Signal and Image Processing (ICSIP) |
| Journal Citation | pp. 190-194 |
| Number of Pages | 5 |
| Year | 2024 |
| Publisher | IEEE (Institute of Electrical and Electronics Engineers) |
| Place of Publication | China |
| Digital Object Identifier (DOI) | https://doi.org/10.1109/ICSIP61881.2024.10671502 |
| Web Address (URL) of Paper | https://ieeexplore.ieee.org/document/10671502 |
| Web Address (URL) of Conference Proceedings | https://ieeexplore.ieee.org/xpl/conhome/10670700/proceeding |
| Conference/Event | 2024 9th International Conference on Signal and Image Processing (ICSIP) |
| Event Details | 2024 9th International Conference on Signal and Image Processing (ICSIP) Delivery In person Event Date 12 to end of 14 Jul 2024 Event Location Nanjing, China |
| Abstract | In scenarios with multiple speakers, humans can selectively attend to a specific speaker through auditory attention to obtain desired information. Similarly, auditory assistance devices rely on auditory attention detection (AAD) to accomplish this task. Current AAD algorithms face challenges such as low signal-to-noise ratio in EEG signals, susceptibility to noise interference from eye and muscle signals, complex associations between signals of different frequencies and auditory attention, and potential impacts of differences in brain health on frequency domain images. In this paper, we propose a novel multi-scale, multi-plane 3D convolutional neural network. Firstly, under the guidance of spatial attention, features are adequately extracted from EEG frequency domain data from multiple plane directions and scales to mitigate noise interference. Secondly, by utilizing multi-channel grouped convolution to decouple features of each channel while capturing potential associations between different frequency features and auditory attention. Finally, a clustering loss function is employed to make classification scores closer to the clustering centers of the categories, enhancing generalization while avoiding overfitting. Experimental results on two datasets demonstrate that our network outperforms competing networks under different window times, which is beneficial for the development of practical neural-guided hearing devices. |
| Keywords | Auditory attention detection; EEG; Deep learning; Brain-computer interfaces |
| Contains Sensitive Content | Does not contain sensitive content |
| ANZSRC Field of Research 2020 | 461103. Deep learning |
| Public Notes | Files associated with this item cannot be displayed due to copyright restrictions. |
| Byline Affiliations | Anhui University, China |
| School of Mathematics, Physics and Computing |
Permalink -
https://research.usq.edu.au/item/zqzq0/a-multi-plane-decoupled-convolutional-network-for-eeg-based-auditory-attention-detection
9
total views2
total downloads2
views this month0
downloads this month