Multi-View Seizure Classification Based on Attention-Based Adaptive Graph ProbSparse Hybrid Network

Abstract

Epilepsy is a neurological disorder characterised by recurrent seizures due to abnormal neuronal discharges. Seizure detection via EEG signals has progressed, but two main challenges are still encountered. First, EEG data can be distorted by physiological factors and external variables, resulting in noisy brain networks. Static adjacency matrices are typically used in current mainstream methods, which neglect the need for dynamic updates and feature refinement. The second challenge stems from the strong reliance on long-range dependencies through self-attention in current methods, which can introduce redundant noise and increase computational complexity, especially in long-duration data. To address these challenges, the Attention-based Adaptive Graph ProbSparse Hybrid Network (AA-GPHN) is proposed. Brain network structures are dynamically optimised using variational inference and the information bottleneck principle, refining the adjacency matrix for improved epilepsy classification. A Linear Graph Convolutional Network (LGCN) is incorporated to focus on first-order neighbours, minimising the aggregation of distant information. Furthermore, a ProbSparse attention-based Informer (PAT) is introduced to adaptively filter long-range dependencies, enhancing efficiency. A joint optimisation loss function is applied to improve robustness in noisy environments. Experimental results on both patient-specific and cross-subject datasets demonstrate that AA-GPHN out-performs existing methods in seizure detection, showing superior effectiveness and generalisation.

Keywords

bioinformatics / deep learning / dynamically / EEG / electroencephalography / ProbSparse attention / seizure classification

References

[1] P. Evangelia, Z. Evangelia, I. Mporas, et al., “Improving Classification of Epileptic and Non-Epileptic EEG Events by Feature Selection,” Neurocomputing 171, no. C (2016): 576-585, https://doi.org/10.1016/j.neucom.2015.06.071.

[2] H. Wang, L. Xu, A. Bezerianos, C. Chen, and Z. Zhang, “Linking Attention-Based Multiscale CNN With Dynamical GCN for Driving Fatigue Detection,” IEEE Transactions on Instrumentation and Mea-surement 70 (2020): 1-11, https://doi.org/10.1109/tim.2020.3047502.

[3] H. Wang, Z. Yuan, H. Zhang, F. Wan, Y. Li, and T. Xu, “Hybrid EEG-fNIRS Decoding With Dynamic Graph Convolutional-Capsule Networks for Motor Imagery/Execution,” Biomedical Signal Processing and Con-trol 104 (2025): 107570, https://doi.org/10.1016/j.bspc.2025.107570.

[4] M. Li, Y. Fang, Y. Wang, et al., “Deep Hypergraph Neural Networks With Tight Framelets,” Proceedings of the AAAI Conference on Artificial Intelligence 39, no. 17 (2025): 18385-18392, https://doi.org/10.1609/aaai.v39i17.34023.

[5] M. Li, Y. Gu, Y. Wang, et al., “When Hypergraph Meets Heterophily: New Benchmark Datasets and Baseline.” Proceedings of the AAAI Conference on Artificial Intelligence 39, no. 17 (2025): 18377-18384, https://doi.org/10.1609/aaai.v39i17.34022.

[6] M. Li, J. Shi, L. Bai, et al., “FrameERC: Framelet Transform Based Multimodal Graph Neural Networks for Emotion Recognition in Con-versation,” Pattern Recognition 161 (2025): 111340, https://doi.org/10.1016/j.patcog.2024.111340.

[7] N. Tishby, F. C. Pereira, and W. Bialek, “The Information Bottleneck Method ” arXiv preprint physics/0004057 2000), https://doi.org/10.48550/arXiv.physics/0004057.

[8] A. A. Alemi, I. Fischer, J. V. Dillon, and K. Murphy, “Deep Varia-tional Information Bottleneck,” arXiv preprint arXiv:1612. 00410 (2016), https://doi.org/10.48550/arXiv.1612.004.

[9] T. Wu, H. Ren, P. Li and J. Leskovec, “Graph Information Bottleneck,” Advances in Neural Information Processing Systems 33 (2020): 20437-20448, https://proceedings.neurips.cc/paper/2020/hash/ebc2aa04e75e3caabda543a1317160c0-Abstract.html.

[10] A. Vaswani, N. Shazeer, N. Parmar, et al., “Attention Is all You Need,” Advances in Neural Information Processing Systems 30 (2017), https://proceedings.neurips.cc/paper/2017/hash/3f5ee243547dee91fbd053c1c4a845aa-Abstract.html.

[11] X. Li, M. Li, P. Yan, et al., “Deep Learning Attention Mechanism in Medical Image Analysis: Basics and Beyonds,” International Journal of Network Dynamics and Intelligence (2023): 93-116, https://doi.org/10.53941/ijndi0201006.

[12] X. Yang,“An Overview of the Attention Mechanisms in Computer Vision,” Journal of Physics: Conference Series 1693, no. 1 (2020): 012173, https://doi.org/10.1088/1742-6596/1693/1/012173.

[13] A. H. Shoeb, Application of Machine Learning to Epileptic Seizure Onset Detection and Treatment (Doctoral dissertation, Massachusetts Institute of Technology, 2009).

[14] Y. Zhao, D. Chu, J. He, et al., “Interactive Local and Global Feature Coupling for EEG-Based Epileptic Seizure Detection,” Biomedical Signal Processing and Control 81 (2023): 104441, https://doi.org/10.1016/j.bspc.2022.104441.

[15] C. Park, G. Choi, J. Kim, et al., “Epileptic Seizure Detection for Multi-Channel EEG With Deep Convolutional Neural Network,” in 2018 International Conference on Electronics, Information, and Communication (ICEIC) IEEE, 2018), 1-5.

[16] M. Zhou, C. Tian, R. Cao, et al., “Epileptic Seizure Detection Based on EEG Signals and CNN,” Frontiers in Neuroinformatics 12 (2018): 95, https://doi.org/10.3389/fninf.2018.00095.

[17] M. Zabihi, S. Kiranyaz, V. Jäntti, T. Lipping, and M. Gabbouj, “Patient-Specific Seizure Detection Using Nonlinear Dynamics and Nullclines,” IEEE Journal of Biomedical and Health Informatics 24, no. 2 (2019): 543-555, https://doi.org/10.1109/JBHI.2019.2906400.

[18] A. Lemkhenter and P. Favaro, Boosting Generalization in Bio-Signal Classification by Learning the Phase-Amplitude Coupling (Springer, 2020),72-85.

[19] X. Hu, S. Yuan, F. Xu, Y. Leng, K. Yuan, and Q. Yuan, “Scalp EEG Classification Using Deep Bi-LSTM Network for Seizure Detection,” Computers in Biology and Medicine 124 (2020): 103919, https://doi.org/10.1016/j.compbiomed.2020.103919.

[20] H. A. Glory, C. Vigneswaran, S. S. Jagtap, R. Shruthi, G. Hariharan, and V. Sriram, “AHW-BGOA-DNN: A Novel Deep Learning Model for Epileptic Seizure Detection,” Neural Computing & Applications 33, no. 11 (2021): 6065-6093, https://doi.org/10.1007/s00521-020-05384-7.

[21] X. Gao, Y. Zhu, Y. Yang, et al., “A Seizure Detection Method Based on Hypergraph Features and Machine Learning,” Biomedical Signal Processing and Control 77 (2022): 103769, https://doi.org/10.1016/j.bspc.2022.103769.

[22] Y. Xu, J. Yang, W. Ming, et al., “Shorter Latency of Real-Time Epileptic Seizure Detection Via Probabilistic Prediction,” Expert Sys-tems with Applications 236 (2024): 121359, https://doi.org/10.1016/j.eswa.2023.121359.

[23] R. Zhu, W. Pan, J. Liu, and J. L. Shang, “Epileptic Seizure Prediction via Multidimensional Transformer and Recurrent Neural Network Fusion,” Journal of Translational Medicine 22, no. 1 (2024): 895, https://doi.org/10.1186/s12967-024-05678-7.

[24] L. Yingjian, L. Guoyang, W. Shibin, and T. Chung, “Phase Spec-trogram of EEG From S-Transform Enhances Epileptic Seizure Detec-tion,” Expert Systems with Applications 262 (2025): 125621, https://doi.org/10.1016/j.eswa.2024.125621.

[25] J. Zhou, L. Liu, Y. Leng, et al., “Both Cross-Patient and Patient-Specific Seizure Detection Based on Self-Organizing Fuzzy Logic,” In-ternational Journal of Neural Systems 32, no. 6 (2022): 2250017, https://doi.org/10.1142/s0129065722500174.

[26] Y. Zhao, J. He, F. Zhu, et al., “Hybrid Attention Network for Epileptic EEG Classification,” International Journal of Neural Systems 33, no. 6 (2023): 2350031, https://doi.org/10.1142/s0129065723500314.

[27] B. Salafian, E. F. Ben-Knaan, N. Shlezinger, S. De Ribaupierre, and N. Farsad, “MICAL: Mutual Information-Based CNN-Aided Learned Factor Graphs for Seizure Detection From EEG Signals,” IEEE Access 11 (2023): 23085-23096, https://doi.org/10.1109/access.2023.3252897.

[28] Z. Wang, M. R. Sperling, D. Wyeth, and A. Guez, “Automated Seizure Detection Based on State-Space Model Identification,” Sensors 24, no. 6 (2024): 1902, https://doi.org/10.3390/s24061902.

[29] S. Y. Shah, H. Larijani, R. M. Gibson, and D. Liarokapis, “Epileptic Seizure Classification Based on Random Neural Networks Using Discrete Wavelet Transform for Electroencephalogram Signal Decom-position,” Applied Sciences 14, no. 2 (2024): 599, https://doi.org/10.3390/app14020599.

[30] P. Zhu, W. Zhou, C. Cao, G. Liu, Z. Liu, and W. Shang, “A Novel SE-TCN-BiGRU Hybrid Network for Automatic Seizure Detection,” IEEE Access 12 (2024): 127328-127340, https://doi.org/10.1109/access.2024.3406909.

[31] T. Dissanayake, T. Fernando, S. Denman, S. Sridharan, and C. Fookes, “Geometric Deep Learning for Subject Independent Epileptic Seizure Prediction Using Scalp EEG Signals,” IEEE Journal of Biomed-ical and Health Informatics 26, no. 2 (2021): 527-538, https://doi.org/10.1109/jbhi.2021.3100297.

[32] A. Pandey, S. K. Singh, S. S. Udmale, and K. K. Shukla, “Epileptic Seizure Classification Using Battle Royale Search and Rescue Optimization-Based Deep LSTM,” IEEE Journal of Biomedical and Health Informatics 26, no. 11 (2022): 5494-5505, https://doi.org/10.1109/jbhi.2022.3203454.

[33] H. K. Fatlawi and A. Kiss, “Similarity-Based Adaptive Window for Improving Classification of Epileptic Seizures With Imbalance EEG Data Stream,” Entropy 24, no. 11 (2022): 1641, https://doi.org/10.3390/e24111641.

[34] L. Jiang, J. He, H. Pan, D. Wu, T. Jiang, and J. Liu, “Seizure Detection Algorithm Based on Improved Functional Brain Network Structure Feature Extraction,” Biomedical Signal Processing and Control 79 (2023): 104053, https://doi.org/10.1016/j.bspc.2022.104053.

[35] Z. Wang, F. Liu, S. Shi, et al., “Automatic Epileptic Seizure Detec-tion Based on Persistent Homology,” Frontiers in Physiology 14 (2023): 1227952, https://doi.org/10.3389/fphys.2023.1227952.

[36] Z. Yanna,D. Changxu, and Z. Gaobo, “EEG-Based Seizure Detec-tion Using Linear Graph Convolution Network With Focal Loss,” Computer Methods and Programs in Biomedicine (2021): 208, https://doi.org/10.1016/j.cmpb.2021.106277.

[37] C. Dong, Y. Zhao, G. Zhang, et al., “Attention-Based Graph ResNet With Focal Loss for Epileptic Seizure Detection,” Journal of Ambient Intelligence and Smart Environments 14, no. 1 (2022): 61-73, https://doi.org/10.3233/ais-210086.

[38] Z. Yongfeng, X. Tiantian, Z. Ziwei, et al., “Hybrid Network for Patient-Specific Seizure Prediction From EEG Data,” Journal of Ambient Intelligence and Smart Environments 33, no. 11 (2023): 2350056, https://doi.org/10.1142/s0129065723500569.

[39] X. Tiantian, W. Ziwei, Y. Zhao, et al., “Self-Supervised Learning With Attention Mechanism for EEG-Based Seizure Detection,” Biomedical Signal Processing and Control 87 (2024): 105464, https://doi.org/10.1016/j.bspc.2023.105464.

[40] H. Lv, S. Wang, H. Feng, et al., “L-Sclnet: A Novel Lightweight Self-Supervised Contrastive Learning Network for Seizure Detection,” Avail-able at SSRN 4960397 2024), http://dx.doi.org/10.2139/ssrn.4960397.

Funding

National Natural Science Foundation of China(U20A20398)

National Natural Science Foundation of China(62076005)

National Natural Science Foundation of China(61906002)

Natural Science Foundation of Anhui Province(2008085MF191)

Natural Science Foundation of Anhui Province(2008085QF306)

University Synergy Innovation Programme of Anhui Province, China(GXXT-2021-002)