基於轉移學習對運動想像的腦電圖多類別分類

以作者查詢圖書館館藏

、以作者查詢臺灣博碩士

、以作者查詢全國書目

、勘誤回報

、線上人數：83

、訪客IP：3.139.97.233

姓名

陳俊宇(Chun-Yu Chen) 查詢紙本館藏

畢業系所

生物醫學工程研究所

論文名稱

基於轉移學習對運動想像的腦電圖多類別分類
(Multiclass Classification of EEG Motor Imagery Signals Based on Transfer Learning)

相關論文

★ 足弓指標參數之比較分析	★ 運用腦電波研究中風病人的復健成效與持續情形
★ 重複間斷性Theta爆發刺激對手部運動之腦波的影響	★ Amylose mediated electricity production of Staphylococcus epidermidis for inhibition of Cutibacterium acnes growth
★ 使用虛擬實境系統誘發事件相關電位P300之研究	★ 虛擬實境誘發體感覺事件相關電位P300之動態因果模型研究
★ 使用GPU提升事件相關電位之動態因果模型的運算效能	★ 基於動態因果模型之老化相關的運動網路研究
★ 應用腦電圖預測中風病人復健情況	★ 以益智遊戲進行空間工作記憶訓練在事件相關電位P3上的影響
★ 基於虛擬實境復健之中風後運動網路功能性重組研究	★ 應用腦電圖與相關臨床因子預測中風病人復原之研究
★ 中風復健後與虛擬實境物理參數相關的動作網絡重組	★ 以運動指標預測復健成效暨設計復健方針
★ 運用時頻轉換分析方法研究工作記憶訓練之人類大腦可塑性	★ 中風患者在復健後的大腦神經連結的變化

檔案

[Endnote RIS 格式]

[Bibtex 格式]

[相關文章]

[文章引用]

[完整記錄]

[館藏目錄]

至系統瀏覽論文 (2027-8-29以後開放)

摘要(中)

轉移學習(Transfer Learning) 是屬於機器學習的一種研究領域，是把已經訓練好的模型參數轉移至新的模型來幫助新模型的訓練，假設大部分數據和任務是存在關聯性的，通過轉移學習可以將模型所學習到的知識透過某種方式來傳遞知識給新模型從而加快並優化模型的學習效率。
EEG 信號的多類分類對於腦機介面 (BCI) 應用至關重要，與兩類案例相比，使用深度學習方法訓練多類模型所需的時間顯著增加。為了加速訓練過程，轉移學習被用來調整從先前訓練的模型中提取的模型參數，以幫助新模型的訓練。然而，由於 EEG 特徵的顯著個體間差異，使用轉移學習對預測模型的泛化受到限制。
本研究共召集4位受試者參與實驗，通過5種運動想像(Motor Imagery)對受試者進行視覺刺激並收集他們的腦電訊號，我們收集了14個電極的時域和頻域訊號，共做了700 次試驗（即每種方向 140 次試驗），並提取他們腦對於五個不同運動方向的特徵，並通過這些特徵進入深度學習分類器進行5類分類。
在這項研究中，我們測試了一種轉移學習方法的信效度，該方法可用於深度學習方法中，將在五個不同運動方向的運動圖像中對多類腦電圖進行快速分類。測量的腦電數據經過標準程序提取時間特徵的ERP和頻率特徵的ERSP，通過轉移學習將模型的泛化能力轉移給其他受試者。通過結果表明，一維頻域和時域特徵在無使用轉移學習的情況Subject-Dependent的準確率分別可以達到88.27%和87.26%，在Subject-Independent的準確率分別可以達到80.67% 和 79.11%。使用二維頻率特徵加上時域特徵在無使用轉移學習的情況Subject-Dependent的準確率最高準確率為86.88%，Subject-Independent的最高準確率則是80.64%。最後使用二維頻率特徵加時域特徵進行轉移學習時，Subject-Dependent的準確率降至73.52%，Subject-Independent的準確率提升至85.86% (訓練時間從110分鐘降至29分鐘)。我們的結果表明Subject-Independent在腦電波多分類上相較於Subject-Dependent會更加困難，並證明對腦電圖多分類使用轉移學習確實能在不降低太多準確率的情況下也減少模型訓練時間。

摘要(英)

Transfer Learning is a research field of machine learning, which is to transfer the trained model parameters to a new model to help build up the new model. It is assumed that most of the data and tasks are related. The knowledge learned from the pre-trained model can be transferred into the new model in some ways to speed up and optimize the learning efficiency of the model.
Multi-class classification of EEG signals is critical for brain-computer interface (BCI) applications, and the time required to train multi-class models using deep learning methods increases significantly compared to the two-class case. Therefore, transfer learning is used to adjust model parameters extracted from previously trained models to aid the training of new models. However, the generalization of predictive models using transfer learning is limited due to the significant inter-subject variability in EEG features.
A total of 4 subjects were recruited to participate in this study. The subjects were visually stimulated through five types of motor imagery and their EEG signals were collected. We collected time and frequency domain signals from 14 electrodes. A total of 700 trials (i.e., 140 trials in each direction) were conducted to extract brain features in 5 different motion directions, and these features were input into a deep learning classifier for 5 categories of classification.
In this study, we tested the reliability of a transfer learning method that can be used in a deep learning approach to rapidly classify multiple types of EEG images in five different motion directions. The measured EEG data undergo a standard procedure to extract ERP for temporal features and ERSP for frequency features, and the generalization ability of the model is transferred to other subjects through transfer learning. The results show that the accuracy of the 1D frequency and time domain features can reach 88.27% and 87.26% for Subject-Dependent without transfer learning, and 80.67% and 79.11% for Subject-Independent. The highest accuracy of 86.88% for Subject-Dependent and 80.64% for Subject-Independent without transfer learning is achieved using 2D frequency features plus time domain features. Finally, when transfer learning was performed using 2D frequency features plus time-domain features, the accuracy of Subject-Dependent decreased to 73.52% and that of Subject-Independent increased to 85.86% (the training time decreased from 110 minutes to 29 minutes). Our results show that Subject-Independent is more difficult than Subject-Dependent in EEG multi categorization, and demonstrate that using transfer learning for EEG multi categorization can indeed reduce the model training time without much reduction in accuracy.

關鍵字(中)

★ 腦電圖
★ 深度學習
★ 轉移學習
★ 運動想像
★ 多類別分類

關鍵字(英)

★ EEG
★ Deep Learning
★ Transfer Learning
★ Motor Imagery
★ Multi-category classification

論文目次

摘要 i
Abstract ii
致謝 iv
目錄 v
圖目錄 vi
表目錄 vii
第一章緒論 1
1.1 腦電波 1
1.1.1 腦波的量測 1
1.1.1 腦波的多分類 2
1.2人工智慧方法與多類別分類 4
1.2.1 傳統機器學習與多分類器 4
1.2.2 深度學習 6
1.2.3 轉移學習 7
1.3 研究背景與動機 9
第二章實驗設計與研究方法 10
2.1 實驗設計 10
2.2 實驗儀器介紹 10
2.3實驗流程與資料前處理 12
2.4 研究資料分析方法 17
2.5 深度學習模型建構 18
2.6 轉移方法 23
第三章研究結果 25
3.1分析方法準確率 25
3.2轉移學習成效 29
第四章討論與結論 31
4.1準確率表現討論 31
4.2提取特徵差異討論 32
4.3轉移效果討論 34
4.4結論 36
第五章未來展望 38
文獻參考 39

參考文獻

[1] D. P. Subha, P. K. Joseph, R. Acharya U, and C. M. Lim, "EEG signal analysis: a survey," Journal of medical systems, vol. 34, no. 2, pp. 195-212, 2010.
[2] S. Sanei and J. A. Chambers, EEG signal processing. John Wiley & Sons, 2013.
[3] J. Malmivuo and R. Plonsey, Bioelectromagnetism: principles and applications of bioelectric and biomagnetic fields. Oxford University Press, USA, 1995.
[4] E. Y. Lew, R. Chavarriaga, S. Silvoni, and J. d. R. Millán, "Single trial prediction of self-paced reaching directions from EEG signals," Frontiers in neuroscience, vol. 8, p. 222, 2014.
[5] M. Teplan, "Fundamentals of EEG measurement," Measurement science review, vol. 2, no. 2, pp. 1-11, 2002.
[6] B. R. Cahn and J. Polich, "Meditation states and traits: EEG, ERP, and neuroimaging studies," Psychological bulletin, vol. 132, no. 2, p. 180, 2006.
[7] R. İnce, S. S. Adanır, and F. Sevmez, "The inventor of electroencephalography (EEG): Hans Berger (1873–1941)," Child′s Nervous System, vol. 37, no. 9, pp. 2723-2724, 2021.
[8] E. Angelakis, J. F. Lubar, S. Stathopoulou, and J. Kounios, "Peak alpha frequency: an electroencephalographic measure of cognitive preparedness," Clinical Neurophysiology, vol. 115, no. 4, pp. 887-897, 2004.
[9] N. A. Busch and R. VanRullen, "Spontaneous EEG oscillations reveal periodic sampling of visual attention," Proceedings of the National Academy of Sciences, vol. 107, no. 37, pp. 16048-16053, 2010.
[10] S. Hanslmayr, J. Gross, W. Klimesch, and K. L. Shapiro, "The role of alpha oscillations in temporal attention," Brain research reviews, vol. 67, no. 1-2, pp. 331-343, 2011.
[11] W. Klimesch, "EEG alpha and theta oscillations reflect cognitive and memory performance: a review and analysis," Brain research reviews, vol. 29, no. 2-3, pp. 169-195, 1999.
[12] M. Rangaswamy et al., "Beta power in the EEG of alcoholics," Biological psychiatry, vol. 52, no. 8, pp. 831-842, 2002.
[13] G. Pfurtscheller and F. L. Da Silva, "Event-related EEG/MEG synchronization and desynchronization: basic principles," Clinical neurophysiology, vol. 110, no. 11, pp. 1842-1857, 1999.
[14] D. A. McCormick, M. J. McGinley, and D. B. Salkoff, "Brain state dependent activity in the cortex and thalamus," Current opinion in neurobiology, vol. 31, pp. 133-140, 2015.
[15] J. Baumeister, T. Barthel, K.-R. Geiss, and M. Weiss, "Influence of phosphatidylserine on cognitive performance and cortical activity after induced stress," Nutritional neuroscience, vol. 11, no. 3, pp. 103-110, 2008.
[16] S. J. Luck, An introduction to the event-related potential technique. MIT press, 2014.
[17] S. Sur and V. K. Sinha, "Event-related potential: An overview," Industrial psychiatry journal, vol. 18, no. 1, p. 70, 2009.
[18] T. W. Picton, "The P300 wave of the human event-related potential," Journal of clinical neurophysiology, vol. 9, no. 4, pp. 456-479, 1992.
[19] T. W. Picton and S. A. Hillyard, "Human auditory evoked potentials. II: Effects of attention," Electroencephalography and clinical neurophysiology, vol. 36, pp. 191-200, 1974.
[20] M. Lotze and U. Halsband, "Motor imagery," Journal of Physiology-paris, vol. 99, no. 4-6, pp. 386-395, 2006.
[21] R. Chatterjee and T. Bandyopadhyay, "EEG based motor imagery classification using SVM and MLP," in 2016 2nd International Conference on Computational Intelligence and Networks (CINE), 2016, pp. 84-89: IEEE.
[22] E. Dong, C. Li, L. Li, S. Du, A. N. Belkacem, and C. Chen, "Classification of multi-class motor imagery with a novel hierarchical SVM algorithm for brain–computer interfaces," Medical & biological engineering & computing, vol. 55, no. 10, pp. 1809-1818, 2017.
[23] M. M. O. Rashid and M. Ahmad, "Multiclass motor imagery classification for BCI application," in 2016 International Workshop on Computational Intelligence (IWCI), 2016, pp. 35-40: IEEE.
[24] Y. Song, D. Wang, K. Yue, N. Zheng, and Z.-J. M. Shen, "EEG-based motor imagery classification with deep multi-task learning," in 2019 International Joint Conference on Neural Networks (IJCNN), 2019, pp. 1-8: IEEE.
[25] S. R. Carvalho et al., "A deep learning approach for classification of reaching targets from EEG images," in 2017 30th SIBGRAPI Conference on Graphics, Patterns and Images (SIBGRAPI), 2017, pp. 178-184: IEEE.
[26] Y. LeCun, Y. Bengio, and G. Hinton, "Deep learning," nature, vol. 521, no. 7553, pp. 436-444, 2015.
[27] P. Cunningham, M. Cord, and S. J. Delany, "Supervised learning," in Machine learning techniques for multimedia: Springer, 2008, pp. 21-49.
[28] Z. Ghahramani, "Unsupervised learning," in Summer school on machine learning, 2003, pp. 72-112: Springer.
[29] M. A. Wiering and M. Van Otterlo, "Reinforcement learning," Adaptation, learning, and optimization, vol. 12, no. 3, p. 729, 2012.
[30] D. Silver et al., "Mastering the game of Go with deep neural networks and tree search," nature, vol. 529, no. 7587, pp. 484-489, 2016.
[31] C. J. Watkins and P. Dayan, "Q-learning," Machine learning, vol. 8, no. 3, pp. 279-292, 1992.
[32] A. R. A. Raziff, M. N. Sulaiman, N. Mustapha, and T. Perumal, "Single classifier, OvO, OvA and RCC multiclass classification method in handheld based smartphone gait identification," in AIP Conference Proceedings, 2017, vol. 1891, no. 1, p. 020009: AIP Publishing LLC.
[33] A. Kamilaris and F. X. Prenafeta-Boldú, "Deep learning in agriculture: A survey," Computers and electronics in agriculture, vol. 147, pp. 70-90, 2018.
[34] A. Bamdadian, C. Guan, K. K. Ang, and J. Xu, "Improving session-to-session transfer performance of motor imagery-based BCI using adaptive extreme learning machine," in 2013 35th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), 2013, pp. 2188-2191: IEEE.
[35] P. Zanini, M. Congedo, C. Jutten, S. Said, and Y. Berthoumieu, "Transfer learning: A Riemannian geometry framework with applications to brain–computer interfaces," IEEE Transactions on Biomedical Engineering, vol. 65, no. 5, pp. 1107-1116, 2017.
[36] F. Fahimi, Z. Zhang, W. B. Goh, T.-S. Lee, K. K. Ang, and C. Guan, "Inter-subject transfer learning with an end-to-end deep convolutional neural network for EEG-based BCI," Journal of neural engineering, vol. 16, no. 2, p. 026007, 2019.
[37] S. J. Pan and Q. Yang, "A survey on transfer learning," IEEE Transactions on knowledge and data engineering, vol. 22, no. 10, pp. 1345-1359, 2009.
[38] S. P. Levine et al., "A direct brain interface based on event-related potentials," IEEE Transactions on Rehabilitation Engineering, vol. 8, no. 2, pp. 180-185, 2000.
[39] A. Schlögl, C. Keinrath, D. Zimmermann, R. Scherer, R. Leeb, and G. Pfurtscheller, "A fully automated correction method of EOG artifacts in EEG recordings," Clinical neurophysiology, vol. 118, no. 1, pp. 98-104, 2007.
[40] M. Sifuzzaman, M. R. Islam, and M. Ali, "Application of wavelet transform and its advantages compared to Fourier transform," 2009.
[41] J. Lin and L. Qu, "Feature extraction based on Morlet wavelet and its application for mechanical fault diagnosis," Journal of sound and vibration, vol. 234, no. 1, pp. 135-148, 2000.
[42] O.-Y. Kwon, M.-H. Lee, C. Guan, and S.-W. Lee, "Subject-independent brain–computer interfaces based on deep convolutional neural networks," IEEE transactions on neural networks and learning systems, vol. 31, no. 10, pp. 3839-3852, 2019.
[43] F. Lotte, C. Guan, and K. K. Ang, "Comparison of designs towards a subject-independent brain-computer interface based on motor imagery," in 2009 Annual International Conference of the IEEE Engineering in Medicine and Biology Society, 2009, pp. 4543-4546: IEEE.
[44] T. Fushiki, "Estimation of prediction error by using K-fold cross-validation," Statistics and Computing, vol. 21, no. 2, pp. 137-146, 2011.
[45] K. Simonyan and A. Zisserman, "Very deep convolutional networks for large-scale image recognition," arXiv preprint arXiv:1409.1556, 2014.
[46] C. Szegedy et al., "Going deeper with convolutions," in Proceedings of the IEEE conference on computer vision and pattern recognition, 2015, pp. 1-9.
[47] S. Targ, D. Almeida, and K. Lyman, "Resnet in resnet: Generalizing residual architectures," arXiv preprint arXiv:1603.08029, 2016.
[48] S. Ioffe and C. Szegedy, "Batch normalization: Accelerating deep network training by reducing internal covariate shift," in International conference on machine learning, 2015, pp. 448-456: PMLR.
[49] K. He, X. Zhang, S. Ren, and J. Sun, "Deep residual learning for image recognition," in Proceedings of the IEEE conference on computer vision and pattern recognition, 2016, pp. 770-778.
[50] I. Kandel and M. Castelli, "How deeply to fine-tune a convolutional neural network: a case study using a histopathology dataset," Applied Sciences, vol. 10, no. 10, p. 3359, 2020.
[51] K. Zhang, N. Robinson, S.-W. Lee, and C. Guan, "Adaptive transfer learning for EEG motor imagery classification with deep Convolutional Neural Network," Neural Networks, vol. 136, pp. 1-10, 2021.
[52] M.-A. Li and D.-Q. Xu, "A Transfer Learning Method based on VGG-16 Convolutional Neural Network for MI Classification," in 2021 33rd Chinese Control and Decision Conference (CCDC), 2021, pp. 5430-5435: IEEE.
[53] N. Shajil, M. Sasikala, and A. Arunnagiri, "Deep learning classification of two-class motor imagery EEG signals using transfer learning," in 2020 International Conference on e-Health and Bioengineering (EHB), 2020, pp. 1-4: IEEE.
[54] J. Li, Y. Wang, L. Zhang, and T.-P. Jung, "Combining ERPs and EEG spectral features for decoding intended movement direction," in 2012 Annual International Conference of the IEEE Engineering in Medicine and Biology Society, 2012, pp. 1769-1772: IEEE.
[55] M. Wei, R. Yang, and M. Huang, "Motor imagery EEG signal classification based on deep transfer learning," in 2021 IEEE 34th International Symposium on Computer-Based Medical Systems (CBMS), 2021, pp. 85-90: IEEE.
[56] M. Arvaneh, C. Guan, K. K. Ang, and C. Quek, "Optimizing the channel selection and classification accuracy in EEG-based BCI," IEEE Transactions on Biomedical Engineering, vol. 58, no. 6, pp. 1865-1873, 2011.
[57] Y. Wang and S. Makeig, "Predicting intended movement direction using EEG from human posterior parietal cortex," in International Conference on Foundations of Augmented Cognition, 2009, pp. 437-446: Springer.
[58] Z. Wang, S. Lyu, G. Schalk, and Q. Ji, "Deep feature learning using target priors with applications in ECoG signal decoding for BCI," in Twenty-Third International Joint Conference on Artificial Intelligence, 2013.
[59] S. Vaid, P. Singh, and C. Kaur, "EEG signal analysis for BCI interface: A review," in 2015 fifth international conference on advanced computing & communication technologies, 2015, pp. 143-147: IEEE.

指導教授

陳純娟(Chun-Chuan Chen)

審核日期

2022-8-29

推文