近年來表面肌電訊號的應用文獻越來越突出，然而在深度學習演算法的發展也不容忽視，因此，基於表面肌電訊號的研究領域中，越來越多的研究採用人工智慧的演算法。但是，許多的研究都無法達到比較好的結果，主要的原因可能是深度學習演算法需要大量的訓練資料，然而表面肌電訊號的資料無法透過一個受試者生成數以萬計的訓練資料來訓練深度學習模型。在此論文中提出結合三維捲積網路及長短期記憶網路的深度學習模型，並提出針對表面肌電訊號的數據增強演算法。同時我們採用兩個訓練資料集來評估實驗結果，第一個是線上開源的資料集，其中包含11位膝關節異常及11位正常的受試者，此資料集主要拿來比較不同神經網路架構的準確率，第二個為自行招募受試者之資料集，採用兩種不同的表面肌電訊號感測器進行量測，分別收集了28位及5位下肢正常與異常的受試者，並透過這兩組資料進行遷移訓練來改善資料量不足導致深度學習模型沒有收斂的問題。實驗結果顯示出我們所提出的方法能有效於辨識基於表面肌電訊號量測之下肢異常的受試者，且準確度達94.12%

In recent years, the application of surface electromyography (sEMG) has increasingly more prominent, while the development of deep learning algorithms cannot be ignored. Therefore, within the field of sEMG-based pattern recognition, more and more AI algorithms are employed. However, many results can not demonstrate high performance. This is due in part to a large amount of data required for the deep learning algorithm. In this paper, a deep neural network that combined 3D-convolutional layers and a long short-term memory layer (LSTM) is proposed. Meanwhile, we propose data augmentation methods and a transfer learning algorithm to improve our performance. In this paper, two datasets are used to evaluate experimental results. The first dataset is an open dataset that is comprised of 11 abnormal and 11 normal participants of the lower limb. The second dataset that referred to as the pre-training and target dataset consists of 28 and 5 examples. This proposed method is shown to outperform the other networks in sEMG-based lower limb abnormal recognition. The experiments with 94.12% accuracy show that our method is effective for lower limb abnormal recognition.

[1] M. A. Oskoei and H. Hu, “Myoelectric control systems a survey,” Biomedical Signal Processing and Control, vol. 2, no. 4, pp. 275–294, 2007.
[2] L. J. Hargrove, K. Englehart and B. Hudgins, “A Comparison of Surface and Intra- muscular Myoelectric Signal Classification,” in IEEE Transactions on Biomedical Engineering, vol. 54, no. 5, pp. 847-853, May 2007.
[3] U. Côté-Allard et al., “Deep Learning for Electromyographic Hand Gesture Signal Classification Using Transfer Learning,” in IEEE Transactions on Neural Systems and Rehabilitation Engineering, vol. 27, no. 4, pp. 760-771, April 2019.
[4] V. von Tscharner and V. Valderrabano, “Classification of multi muscle activation patterns of osteoarthritis patients during level walking,” Journal of Electromyogra- phy and Kinesiology, vol. 20, pp. 676-683, 2010/08/01/ 2010.
[5] C. D. Joshi, U. Lahiri and N. V. Thakor, “Classification of gait phases from lower limb EMG: Application to exoskeleton orthosis,” 2013 IEEE Point-of-Care Health- care Technologies (PHT), Bangalore, 2013, pp. 228-231.
39
[6] K.Englehart,B.HudginandP.A.Parker,“Awavelet-basedcontinuousclassification scheme for multifunction myoelectric control,” in IEEE Transactions on Biomedical Engineering, vol. 48, no. 3, pp. 302-311, March 2001.
[7] K. Lee, “EMG-Based Speech Recognition Using Hidden Markov Models With Global Control Variables,” in IEEE Transactions on Biomedical Engineering, vol. 55, no. 3, pp. 930-940, March 2008.
[8] X. Zhai, B. Jelfs, R. H. M. Chan and C. Tin, “Short latency hand movement clas- sification based on surface EMG spectrogram with PCA,” 2016 38th Annual In- ternational Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), Orlando, FL, 2016, pp. 327-330.
[9] K.Alex,S.IlyaandGeoffreyE.Hinton,“ImageNetClassificationwithDeepConvo- lutional Neural Networks,” in Advances in Neural Information Processing Systems. 1097-1105, 2012.
[10] R. Girshick, “Fast R-CNN,” 2015 IEEE International Conference on Computer Vi- sion (ICCV), Santiago, 2015, pp. 1440-1448.
[11] S. Levine, C. Finn, T. Darrell, and P. Abbeel, “End-to-end training of deep visuo- motor policies,” J. Mach. Learn. Res., vol. 17, no. 39, pp. 1–40, 2016.
[12] W. Feng, N. Guan, Y. Li, X. Zhang and Z. Luo, “Audio visual speech recognition with multimodal recurrent neural networks,” 2017 International Joint Conference on Neural Networks (IJCNN), Anchorage, AK, 2017, pp. 681-688.
[13] A. Jettakul, C. Thamjarat, K. Liaowongphuthorn, C. Udomcharoenchaikit, P. Va- teekul and P. Boonkwan, “A Comparative Study on Various Deep Learning Techniques for Thai NLP Lexical and Syntactic Tasks on Noisy Data,” 2018 15th Inter- national Joint Conference on Computer Science and Software Engineering (JCSSE), Nakhonpathom, 2018, pp. 1-6.
[14] Y. Wu, M. Schuster, Z. Chen, Q. V. Le, M. Norouzi, W. Macherey, M. Krikun, Y. Cao, Q. Gao, K. Macherey, J. Klingner, A. Shah, M. Johnson, X. Liu, Ł. Kaiser, S. Gouws, Y. Kato, T. Kudo, H. Kazawa, K. Stevens, G. Kurian, N. Patil, W. Wang, C. Young, J. Smith, J. Riesa, A. Rudnick, O. Vinyals, G. Corrado, M. Hughes, J. Dean, “Google’s Neural Machine Translation System: Bridging the Gap between Human and Machine Translation,” CoRR, abs/1609.08144, 2016.
[15] U. Côté Allard et al., “A convolutional neural network for robotic arm guidance using sEMG based frequency-features,” 2016 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), Daejeon, 2016, pp. 2464-2470.
[16] Y.YamanoiandR.Kato,“Controlmethodformyoelectrichandusingconvolutional neural network to simplify learning of EMG signals,” 2017 IEEE International Con- ference on Cyborg and Bionic Systems (CBS), Beijing, 2017, pp. 114-118.
[17] F. Quivira, T. Koike-Akino, Y. Wang and D. Erdogmus, “Translating sEMG signals to continuous hand poses using recurrent neural networks,” 2018 IEEE EMBS Inter- national Conference on Biomedical and Health Informatics (BHI), Las Vegas, NV, 2018, pp. 166-169.
[18] J. Sensinger, B. Lock, and T. Kuiken, “Adaptive pattern recognition of myoelec- tric signals: Exploration of conceptual framework and practical algorithms,” IEEE Trans. Neural Syst. Rehabil. Eng., vol. 17, no. 3, pp. 270-278, Jun. 2009.
[19] C. Castellini, A. E. Fiorilla, and G. Sandini, “Multi-subject/daily-life activity emg- based control of mechanical hands,” Journal of neuroengineering and rehabilitation, vol. 6, no. 1, p. 41, 2009.
[20] T.Tommasi,F.Orabona,C.Castellini,andB.Caputo,“Improvingcontrolofdexter- ous hand prostheses using adaptive learning,” IEEE Transactions on Robotics, vol. 29, no. 1, pp. 207–219, 2013.
[21] N. Patricia, T. Tommasit, and B. Caputo, “Multi-source adaptive learning for fast control of prosthetics hand,” in Pattern Recognition (ICPR), 2014 22nd International Conference on. IEEE, 2014, pp. 2769–2774.
[22] M. Heideman, D. Johnson and C. Burrus, “Gauss and the history of the fast fourier transform,” in IEEE ASSP Magazine, vol. 1, no. 4, pp. 14-21, October 1984.
[23] K.Englehart,B.Hudgins,P.A.Parker,andM.Stevenson,“Classificationofthemy- oelectric signal using time-frequency based representations,” Medical engineering and physics, vol. 21, no. 6, pp. 431–438, 1999.
[24] Y.Du,W.Jin,W.Wei,Y.Hu,andW.Geng,“Surfaceemg-basedinter-sessiongesture recognition enhanced by deep domain adaptation,” Sensors, vol. 17, no. 3, p. 458, 2017.
[25] VijilaCKS,KumarCES,“CancellationofECGinelectromyogramusingbackProp- agation network,” in International Conference on advances in recent technologies in communication and computing, Kottayam, Kerala, 630-634. doi: 10.1109/ART- Com.2009.69.
[26] M. Baccouche, F. Mamalet, C. Wolf, C. Garcia, and A. Baskurt, “Sequential deep learning for human action recognition,” in International Workshop on Human Be- havior Understanding. Springer, 2011, pp. 29–39.
[27] A.Karpathy,G.Toderici,S.Shetty,T.Leung,R.Sukthankar,andL.Fei-Fei,“Large- scale video classification with convolutional neural networks,” in Proceedings of the IEEE conference on Computer Vision and Pattern Recognition, 2014, pp. 1725-1732.
[28] N. Lu, Y. Wu, L. Feng and J. Song, “Deep Learning for Fall Detection: 3D-CNN Combined with LSTM on Video Kinematic Data,” in IEEE Journal of Biomedical and Health Informatics, 2018.
[29] D, Dheeru and G, Casey, “UCI Machine Learning Repository,” in University of California, Irvine, School of Information and Computer Sciences, 2017.
[30] C. Goutte and É. Gaussier, “A Probabilistic Interpretation of Precision, Recall and F-Score, with Implication for Evaluation,” ECIR, 2005.
[31] R. Merletti and P. Di Torino, “Standards for reporting emg data,” J Electromyogr Kinesiol, vol. 9, no. 1, pp. 3–4, 1999.