Understanding of Task-Specific and Subject-Specific Components in Surface EMG

Surface electromyogram (sEMG) signals are widely used in human–machine interfaces for gesture recognition and user identification, but existing models often struggle with generalization across different individuals due to subject-specific neuromuscular characteristics. This study introduced a disentanglement model to separate task-specific and subject-specific components from sEMG signals, thus improving the generalization and interpretability of gesture recognition and user identification systems. Experimental results demonstrate that disentangled task-specific components significantly improve the accuracy of both gesture classification and user identification across different subjects and days, outperforming conventional methods in the same scenario. Further analysis of the extracted components reveals that task-specific components capture consistent activation patterns for the same gestures across individuals. In contrast, subject-specific components reflect unique neuromuscular characteristics that can be used for user identification. Notably, subject-specific components show reduced similarity compared to task-specific components in inter-day scenarios, contributing to more accuracy decrease in user identification than in gesture recognition. These findings suggest that the disentanglement approach not only boosts classification performance but also provides deeper insights into the physiological mechanisms underlying sEMG signals. The model’s ability to isolate and interpret different neuromuscular components holds promise for enhancing the robustness of sEMG-based applications in real-world settings, such as rehabilitation and user authentication.