EEG signal classification based on multi-scale sliding-window attention temporal convolutional networks

doi:10.3785/j.issn.1008-973X.2026.02.015

Journal of ZheJiang University (Engineering Science)

2026, Vol. 60

Issue (2): 370-378 DOI: 10.3785/j.issn.1008-973X.2026.02.015

EEG signal classification based on multi-scale sliding-window attention temporal convolutional networks

Xianhua LI1,2(

),Pengfei DU3,Tao SONG4,Xun QIU3,Yu CAI3

1. The First Hospital of Anhui University of Science and Technology, Huainan 232001, China
2. School of Mechatronics Engineering, Anhui University of Science and Technology, Huainan 232001, China
3. School of Artificial Intelligence, Anhui University of Science and Technology, Huainan 232001, China
4. School of Electrical and Mechanical Engineering, Shanghai University, Shanghai 200444, China

Download:

HTML

PDF(3347KB) HTML
Export: BibTeX | EndNote (RIS)

Abstract

A multi-scale sliding-window attention temporal convolutional network (MSWATCN) was proposed to fully exploit the spatio-temporal information of motor imagery electroencephalography (MI-EEG) signals for enhanced classification accuracy. Accurate decoding of the complex spatio-temporal characteristics of MI-EEG signals was achieved by combining multiscale two-stream group convolution, a sliding-window multi-head attention mechanism, and a windowed temporal convolution module. The underlying spatio-temporal features of the signal were first extracted using the multi-scale convolution module, followed by a focus on local key features through the sliding-window attention mechanism to highlight information crucial for classification. The windowed temporal convolution module was employed to model long-term dependencies in the time series, whereby the model’s ability to encode sequential information was significantly improved. Experimental results showed that MSWATCN outperformed all comparison networks and benchmark models in terms of classification accuracy and consistency on the BCI Competition IV 2a and 2b datasets.

Key words： motion imagery multi-scale convolution multi-head attention mechanism sliding window temporal convolutional network

Received: 25 February 2025 Published: 03 February 2026

CLC:

TP 391

Fund: 安徽省重点研究与开发计划项目（2022i01020015）；安徽理工大学医学专项项目（YZ2023H2B013）.

	Service
	E-mail this article
	Add to my bookshelf
	Add to citation manager
	E-mail Alert
	RSS
	Articles by authors
	Xianhua LI
	Pengfei DU
	Tao SONG
	Xun QIU
	Yu CAI

Cite this article:

Xianhua LI,Pengfei DU,Tao SONG,Xun QIU,Yu CAI. EEG signal classification based on multi-scale sliding-window attention temporal convolutional networks. Journal of ZheJiang University (Engineering Science), 2026, 60(2): 370-378.

URL:

https://www.zjujournals.com/eng/10.3785/j.issn.1008-973X.2026.02.015 OR https://www.zjujournals.com/eng/Y2026/V60/I2/370

基于多尺度滑窗注意力时序卷积网络的脑电信号分类

为了提升运动想象脑电（MI-EEG）信号的分类精度，提出多尺度滑窗注意力时序卷积网络（MSWATCN），充分挖掘MI-EEG信号的时空信息. 结合多尺度双流分组卷积、滑动窗口多头注意力机制和窗口化时间卷积模块，实现对MI-EEG信号复杂时空特性的精准解码. 利用多尺度卷积模块提取信号的底层时空特征，通过滑动窗口注意力机制聚焦局部关键特征，突出对分类任务重要的信息. 窗口化时间卷积模块通过建模时间序列中的长期依赖关系，增强模型处理时序信息的能力. 实验结果表明，MSWATCN在BCI Competition IV 2a和2b数据集上的分类准确率和一致性优于对比网络和基准模型.

关键词： 运动想象, 多尺度卷积, 多头注意力机制, 滑动窗口, 时序卷积网络

Fig.1 Overall architecture of multi-scale sliding-window attention temporal convolutional network

Fig.2 Multiscale two-stream group convolution module

Fig.3 Structure of multi-head attention module

Fig.4 Temporal convolutional network module

Tab.1 Classification accuracy of different methods on BCI Competition IV 2a dataset

Tab.2 Classification accuracy of different methods on BCI Competition IV 2b dataset

Fig.5 Structure of multi-scale convolution

Fig.6 Visualization results of t-SNE features of different modules for BCI Competition IV 2a dataset

Fig.7 Visualization results of t-SNE features of different modules for BCI Competition IV 2b dataset

Fig.8 Analysis of relationship between number of sliding windows and classification accuracy

Tab.3 Ablation experiment on modules of multi-scale sliding-window attention temporal convolutional network


[1]	刘近贞, 叶方方, 熊慧基于卷积神经网络的多类运动想象脑电信号识别[J]. 浙江大学学报: 工学版, 2021, 55 (11): 2054- 2066 LIU Jinzhen, YE Fangfang, XIONG Hui Recognition of multi-class motor imagery EEG signals based on convolutional neural network[J]. Journal of Zhejiang University: Engineering Science, 2021, 55 (11): 2054- 2066

[2]	WANG X, LIESAPUTRA V, LIU Z, et al An in-depth survey on deep learning-based motor imagery electroencephalogram (EEG) classification[J]. Artificial Intelligence in Medicine, 2024, 147: 102738 doi: 10.1016/j.artmed.2023.102738

[3]	METZGER S L, LITTLEJOHN K T, SILVA A B, et al A high-performance neuroprosthesis for speech decoding and avatar control[J]. Nature, 2023, 620 (7976): 1037- 1046 doi: 10.1038/s41586-023-06443-4

[4]	WILLETT F R, KUNZ E M, FAN C, et al A high-performance speech neuroprosthesis[J]. Nature, 2023, 620 (7976): 1031- 1036 doi: 10.1038/s41586-023-06377-x

[5]	YOU Y, CHEN W, ZHANG T Motor imagery EEG classification based on flexible analytic wavelet transform[J]. Biomedical Signal Processing and Control, 2020, 62: 102069 doi: 10.1016/j.bspc.2020.102069

[6]	ANOWAR F, SADAOUI S, SELIM B Conceptual and empirical comparison of dimensionality reduction algorithms (PCA, KPCA, LDA, MDS, SVD, LLE, ISOMAP, LE, ICA, t-SNE)[J]. Computer Science Review, 2021, 40: 100378 doi: 10.1016/j.cosrev.2021.100378

[7]	THOMAS K P, GUAN C, LAU C T, et al A new discriminative common spatial pattern method for motor imagery brain–computer interfaces[J]. IEEE Transactions on Biomedical Engineering, 2009, 56 (11): 2730- 2733 doi: 10.1109/TBME.2009.2026181

[8]	SAKHAVI S, GUAN C, YAN S Learning temporal information for brain-computer interface using convolutional neural networks[J]. IEEE Transactions on Neural Networks and Learning Systems, 2018, 29 (11): 5619- 5629 doi: 10.1109/TNNLS.2018.2789927

[9]	SCHIRRMEISTER R T, SPRINGENBERG J T, FIEDERER L D J, et al Deep learning with convolutional neural networks for EEG decoding and visualization[J]. Human Brain Mapping, 2017, 38 (11): 5391- 5420 doi: 10.1002/hbm.23730

[10]	LAWHERN V J, SOLON A J, WAYTOWICH N R, et al EEGNet: a compact convolutional neural network for EEG-based brain–computer interfaces[J]. Journal of Neural Engineering, 2018, 15 (5): 056013 doi: 10.1088/1741-2552/aace8c

[11]	INGOLFSSON T M, HERSCHE M, WANG X, et al. EEG-TCNet: an accurate temporal convolutional network for embedded motor-imagery brain–machine interfaces [C]// Proceedings of the 2020 IEEE International Conference on Systems, Man, and Cybernetics (SMC). [S.l.]: ACM, 2020: 2958–2965.

[12]	LI D, XU J, WANG J, et al A multi-scale fusion convolutional neural network based on attention mechanism for the visualization analysis of EEG signals decoding[J]. IEEE Transactions on Neural Systems and Rehabilitation Engineering, 2020, 28 (12): 2615- 2626 doi: 10.1109/TNSRE.2020.3037326

[13]	ALTAHERI H, MUHAMMAD G, ALSULAIMAN M Dynamic convolution with multilevel attention for EEG-based motor imagery decoding[J]. IEEE Internet of Things Journal, 2023, 10 (21): 18579- 18588 doi: 10.1109/JIOT.2023.3281911

[14]	HAN J W, BAK S, KIM J M, et al META-EEG: meta-learning-based class-relevant EEG representation learning for zero-calibration brain–computer interfaces[J]. Expert Systems with Applications, 2024, 238: 121986 doi: 10.1016/j.eswa.2023.121986

[15]	ZHANG Y, QIU S, WEI W, et al Dynamic weighted filter bank domain adaptation for motor imagery brain–computer interfaces[J]. IEEE Transactions on Cognitive and Developmental Systems, 2023, 15 (3): 1348- 1359 doi: 10.1109/TCDS.2022.3209801

[16]	VASWANI A, SHAZEER N, PARMAR N, et al. Attention is all you need [C]// Proceedings of the 31st International Conference on Neural Information Processing Systems. Long Beach: [s.n.], 2017: 5998–6008.

[17]	MUSALLAM Y K, ALFASSAM N I, MUHAMMAD G, et al Electroencephalography-based motor imagery classification using temporal convolutional network fusion[J]. Biomedical Signal Processing and Control, 2021, 69: 102826 doi: 10.1016/j.bspc.2021.102826

[18]	DING Y F, JIA M P, MIAO Q H, et al A novel time–frequency transformer based on self–attention mechanism and its application in fault diagnosis of rolling bearings[J]. Mechanical Systems and Signal Processing, 2022, 168: 108616 doi: 10.1016/j.ymssp.2021.108616

[19]	ALTAHERI H, MUHAMMAD G, ALSULAIMAN M Physics-informed attention temporal convolutional network for EEG-based motor imagery classification[J]. IEEE Transactions on Industrial Informatics, 2023, 19 (2): 2249- 2258 doi: 10.1109/TII.2022.3197419

[20]	施锦河, 沈继忠, 王攀四类运动想象脑电信号特征提取与分类算法[J]. 浙江大学学报: 工学版, 2012, 46 (2): 338- 344 SHI Jinhe, SHEN Jizhong, WANG Pan Feature extraction and classification of four-class motor imagery EEG data[J]. Journal of Zhejiang University: Engineering Science, 2012, 46 (2): 338- 344

[21]	HE H, WU D Transfer learning for brain–computer interfaces: a euclidean space data alignment approach[J]. IEEE Transactions on Biomedical Engineering, 2020, 67 (2): 399- 410 doi: 10.1109/TBME.2019.2913914

[22]	ANG K K, CHIN Z Y, ZHANG H, et al. Filter bank common spatial pattern (FBCSP) in brain-computer interface [C]// Proceedings of the IEEE International Joint Conference on Neural Networks. Hong Kong: IEEE, 2008: 2390–2397.

[23]	ZHAO H, ZHENG Q, MA K, et al Deep representation-based domain adaptation for nonstationary EEG classification[J]. IEEE Transactions on Neural Networks and Learning Systems, 2021, 32 (2): 535- 545 doi: 10.1109/TNNLS.2020.3010780

[24]	CAO J, LI G, SHEN J, et al IFBCLNet: spatio-temporal frequency feature extraction-based MI-EEG classification convolutional network[J]. Biomedical Signal Processing and Control, 2024, 92: 106092 doi: 10.1016/j.bspc.2024.106092

[25]	SONG Y, ZHENG Q, LIU B, et al EEG conformer: convolutional transformer for EEG decoding and visualization[J]. IEEE Transactions on Neural Systems and Rehabilitation Engineering, 2023, 31: 710- 719 doi: 10.1109/TNSRE.2022.3230250

[26]	吴书晗, 王丹, 陈远方, 等融合注意力的滤波器组双视图图卷积运动想象脑电分类[J]. 浙江大学学报: 工学版, 2024, 58 (7): 1326- 1335 WU Shuhan, WANG Dan, CHEN Yuanfang, et al Attention-fused filter bank dual-view graph convolution motor imagery EEG classification[J]. Journal of Zhejiang University: Engineering Science, 2024, 58 (7): 1326- 1335

[1]	Xinjian XIANG,Tianshun YUAN,Yaqiang HE,Chengli WANG. Traffic flow prediction based on time series decomposition and soft thresholding temporal convolution[J]. Journal of ZheJiang University (Engineering Science), 2025, 59(7): 1353-1361.

[2]	Bote WANG,Qing WANG,Qiang LIU,Bo JIN. Self-supervised anomaly recognition method for wind turbine blade based on multi-channel vibration principal features[J]. Journal of ZheJiang University (Engineering Science), 2025, 59(3): 653-660.

[3]	Longxue LIANG,Chenglong HE,Xiaosuo WU,Haowen YAN. Remote sensing image semantic segmentation network based on global information extraction and reconstruction[J]. Journal of ZheJiang University (Engineering Science), 2024, 58(11): 2270-2279.

[4]	Hai-feng LI,Xue-ying ZHANG,Shu-fei DUAN,Hai-rong JIA,Hui-zhi LIANG. Fusing generative adversarial network and temporal convolutional network for Mandarin emotion recognition[J]. Journal of ZheJiang University (Engineering Science), 2023, 57(9): 1865-1875.

[5]	Chao REN,Gao-wei YAN,Lan CHENG,Fang WANG. Transition mode identification method based on maximum mean discrepancy for multimode process[J]. Journal of ZheJiang University (Engineering Science), 2021, 55(3): 563-570.

[6]	CHEN Jia-Gan, HE Yan, JIANG Jing-Ping. Improved FastSLAM algorithm based on importance weight smoothing[J]. Journal of ZheJiang University (Engineering Science), 2010, 44(8): 1454-1459.

Viewed

Full text

Abstract

Cited

Shared

Discussed