Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
158 tokens/sec
GPT-4o
7 tokens/sec
Gemini 2.5 Pro Pro
45 tokens/sec
o3 Pro
4 tokens/sec
GPT-4.1 Pro
38 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

Electromyography Based Cross-Subject Limb Angle Estimation via Hierarchical Spiking Attentional Feature Decomposition Network (2404.07517v2)

Published 11 Apr 2024 in cs.HC

Abstract: As human-machine interaction systems are developing towards lightweight and pervasive direction, the role of simultaneous and proportional control (SPC) in human-machine interaction becomes increasingly prominent. However, existing continuous joint angle prediction algorithms based on surface electromyography (sEMG) typically incur high inference costs or are only applicable to specific subjects rather than cross-subject scenarios. Therefore, we proposed a hierarchical Spiking Attentional FEature decomposition Network (SAFE-Net) in order to reduce inference costs and improve recognition accuracy in cross-subject scenarios. This network first encodes the sEMG signals into neural spiking forms through a Spiking Sparse Attention Encoder (SSAE). The compressed features are then decomposed into kinematic features and biological features by a Spiking Attentional Feature Decomposition (SAFD) module. Finally, the kinematic features and biological features are decoded into joint angle values and subject identity, respectively. We validated the effectiveness of SAFE-Net on two datasets (SIAT-DB1 and SIAT-DB2) and compared it with two state-of-the-art methods, Informer and Spikformer. Experimental results demonstrate that, on the one hand, SSAE saves 39.1% and 37.5% power consumption respectively over them in terms of inference costs. On the other hand, SAFE-Net outperforms Informer and Spikformer in recognition accuracy on both datasets. This study showcased that the proposed SAFE-Net can provide accurate predictions in cross-subject scenarios, offering a promising vision for precise continuous control of lower limb rehabilitation exoskeleton robots.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (45)
  1. An empirical evaluation of generic convolutional and recurrent networks for sequence modeling. arXiv preprint arXiv:1803.01271 .
  2. A review on emg-based motor intention prediction of continuous human upper limb motion for human-robot collaboration. Biomedical Signal Processing and Control 51, 113–127.
  3. A bio-inspired spiking attentional neural network for attentional selection in the listening brain. IEEE Transactions on Neural Networks and Learning Systems .
  4. semg-based continuous estimation of finger kinematics via large-scale temporal convolutional network. Applied Sciences 11, 4678.
  5. Hand gesture recognition based on surface electromyography using convolutional neural network with transfer learning method. IEEE Journal of Biomedical and Health Informatics 25, 1292–1304.
  6. Alignment-enhanced interactive fusion model for complete and incomplete multimodal hand gesture recognition. IEEE Transactions on Neural Systems and Rehabilitation Engineering 31, 4661–4671.
  7. A neuroscience-inspired spiking neural network for eeg-based auditory spatial attention detection. Neural Networks 152, 555–565.
  8. Spikingjelly: An open-source machine learning infrastructure platform for spike-based intelligence. Science Advances 9, eadi1480.
  9. Multi-attention feature fusion network for accurate estimation of finger kinematics from surface electromyographic signals. IEEE Transactions on Human-Machine Systems .
  10. Long short-term memory. Neural computation 9, 1735–1780.
  11. Long time prediction of human lower limb movement based on ipso-bpnn, in: Journal of Physics: Conference Series, IOP Publishing. p. 042099.
  12. A review of the key technologies for semg-based human-robot interaction systems. Biomedical Signal Processing and Control 62, 102074.
  13. A neural network-based model for lower limb continuous estimation against the disturbance of uncertainty. Biomedical Signal Processing and Control 71, 103115.
  14. Gesture recognition using surface electromyography and deep learning for prostheses hand: state-of-the-art, challenges, and future. Frontiers in neuroscience 15, 621885.
  15. Cross-user gesture recognition from semg signals using an optimal transport assisted student-teacher framework. Computers in Biology and Medicine 165, 107327.
  16. Metric learning for robust gait phase recognition for a lower limb exoskeleton robot based on semg. IEEE Transactions on Medical Robotics and Bionics 4, 472–479.
  17. A novel method for detecting misclassifications of the locomotion mode in lower-limb exoskeleton robot control. IEEE Robotics and Automation Letters 7, 7779–7785.
  18. A novel and efficient feature extraction method for deep learning based continuous estimation. IEEE Robotics and Automation Letters 6, 7341–7348.
  19. A bi-directional lstm network for estimating continuous upper limb movement from surface electromyography. IEEE Robotics and Automation Letters 6, 7217–7224.
  20. Continuous estimation of upper limb joint angle from semg signals based on sca-lstm deep learning approach. Biomedical Signal Processing and Control 61, 102024.
  21. Visualizing data using t-sne. Journal of machine learning research 9.
  22. Few-shot learning for decoding surface electromyography for hand gesture recognition, in: ICASSP 2021-2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), IEEE. pp. 1300–1304.
  23. Fs-hgr: Few-shot learning for hand gesture recognition via electromyography. IEEE transactions on neural systems and rehabilitation engineering 29, 1004–1015.
  24. Feature decomposition and reconstruction learning for effective facial expression recognition, in: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pp. 7660–7669.
  25. A review on electromyography decoding and pattern recognition for human-machine interaction. Ieee Access 7, 39564–39582.
  26. A single platform for classification and prediction using a hybrid bioinspired and deep neural network (pso-lstm). Mapan 37, 47–58.
  27. Continuous estimation of human upper limb joint angles by using pso-lstm model. IEEE Access 9, 17986–17997.
  28. A spiking network classifies human semg signals and triggers finger reflexes on a robotic hand. Robotics and Autonomous Systems 131, 103566.
  29. Attention is all you need. Advances in neural information processing systems 30.
  30. A review of gait phase detection algorithms for lower limb prostheses. Sensors 20, 3972.
  31. Prediction of contralateral lower-limb joint angles using vibroarthrography and surface electromyography signals in time-series network. IEEE Transactions on Automation Science and Engineering 20, 901–908.
  32. Similarity function for one-shot learning to enhance the flexibility of myoelectric interfaces. IEEE Transactions on Neural Systems and Rehabilitation Engineering 31, 1697–1706.
  33. Masked spiking transformer, in: Proceedings of the IEEE/CVF International Conference on Computer Vision, pp. 1761–1771.
  34. A tandem learning rule for effective training and rapid inference of deep spiking neural networks. IEEE Transactions on Neural Networks and Learning Systems 34, 446–460.
  35. Progressive tandem learning for pattern recognition with deep spiking neural networks. IEEE Transactions on Pattern Analysis and Machine Intelligence 44, 7824–7840.
  36. Gait phase classification for a lower limb exoskeleton system based on a graph convolutional network model. IEEE Transactions on Industrial Electronics 69, 4999–5008.
  37. Cross-user electromyography pattern recognition based on a novel spatial-temporal graph convolutional network. IEEE Transactions on Neural Systems and Rehabilitation Engineering .
  38. A novel event-driven spiking convolutional neural network for electromyography pattern recognition. IEEE Transactions on Biomedical Engineering .
  39. Spike-driven transformer. Advances in Neural Information Processing Systems 36.
  40. Informer: Beyond efficient transformer for long sequence time-series forecasting, in: Proceedings of the AAAI conference on artificial intelligence, pp. 11106–11115.
  41. Continuous estimation of lower limb joint angles from multi-stream signals based on knowledge tracing. IEEE Robotics and Automation Letters 8, 951–957.
  42. Time–frequency feature transform suite for deep learning-based gesture recognition using semg signals. Robotica 41, 775–788.
  43. Spikformer: When spiking neural network meets transformer. arXiv preprint arXiv:2209.15425 .
  44. Learn-to-decompose: cascaded decomposition network for cross-domain few-shot facial expression recognition, in: European Conference on Computer Vision, Springer. pp. 683–700.
  45. When facial expression recognition meets few-shot learning: A joint and alternate learning framework, in: Proceedings of the AAAI Conference on Artificial Intelligence, pp. 5367–5375.

Summary

We haven't generated a summary for this paper yet.

Ai Sparkles Streamline Icon: https://streamlinehq.com Summarize Now
X Twitter Logo Streamline Icon: https://streamlinehq.com

Tweets