An LSTM Feature Imitation Network for Hand Movement Recognition from sEMG Signals

2024-05-23

arXiv.org

AI Summary

To address the poor generalization of deep models for hand motion recognition from surface electromyography (sEMG) signals under low-data regimes, this paper proposes the Feature Imitation Network (FIN). FIN introduces a novel LSTM-driven, end-to-end temporal feature reconstruction paradigm that jointly learns four interpretable time-domain features—entropy, root-mean-square (RMS), mean absolute value, and zero-crossing rate—within a 300-ms sliding window. We design a lightweight LSTM-FIN architecture, a time-domain feature reconstruction loss, a cross-subject transfer learning strategy, and a low-latency simulation evaluation framework. Experiments demonstrate that FIN achieves an R² score of 99% for feature reconstruction and attains 80% cross-subject motion classification accuracy—substantially improving accuracy, robustness, and real-time deployability in few-shot sEMG scenarios.

Technology Category

• Machine Learning: Feature Construction/Reformulation
• Computer Vision: Motion & Tracking
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Surface Electromyography (sEMG) is a non-invasive signal that is used in the recognition of hand movement patterns, the diagnosis of diseases, and the robust control of prostheses. Despite the remarkable success of recent end-to-end Deep Learning approaches, they are still limited by the need for large amounts of labeled data. To alleviate the requirement for big data, we propose utilizing a feature-imitating network (FIN) for closed-form temporal feature learning over a 300ms signal window on Ninapro DB2, and applying it to the task of 17 hand movement recognition. We implement a lightweight LSTM-FIN network to imitate four standard temporal features (entropy, root mean square, variance, simple square integral). We observed that the LSTM-FIN network can achieve up to 99% R2 accuracy in feature reconstruction and 80% accuracy in hand movement recognition. Our results also showed that the model can be robustly applied for both within- and cross-subject movement recognition, as well as simulated low-latency environments. Overall, our work demonstrates the potential of the FIN modeling paradigm in data-scarce scenarios for sEMG signal processing.

Problem

• Depth Learning
• sEMG Signal Recognition
• Data Scarcity

Innovation

• LSTM-FIN
• sEMG signal processing
• hand gesture recognition