EMGBlend improves muscle signal decoding across diverse datasets
EMGBlend: Heterogeneity-Aware Self-Supervised Pretraining for Gesture and Force Decoding
Machine LearningArtificial Intelligence
Summary
Surface electromyography (EMG) devices that read muscle signals come in many shapes with different sensors and qualities. The authors found that simply mixing all these different EMG datasets for training models causes problems and poor results. They created EMGBlend, a new way to train large models that respects each dataset's unique layout and frequency limits, balancing their influence. Their method works better for recognizing hand gestures and estimating muscle force and shows how to combine diverse muscle data smartly.
What this means in practice
- •For wearable device developers: Train models that better interpret muscle signals from varied EMG hardware for more accurate control of wearable tech and prosthetics.$Commercial implications: Enables production of improved muscle-controlled wearable devices by providing robust pretrained models adaptable to diverse sensor setups.
- •For human-computer interface engineers: Integrate pretrained EMG models that handle different sensor layouts to enhance gesture recognition accuracy in novel input systems.
Authors
Yuwei Jia, Cheng Zhong, Jinyang Yu, Zhe Cui
Abstract
Public surface electromyography (EMG) datasets vary widely in electrode layout, channel count, frequency support, and size. Simply mixing them for pretraining can misalign channel semantics, introduce spectral targets that some devices cannot observe, and let large or high-channel-count datasets dominate learning. We introduce EMGBlend, a self-supervised framework designed around these differences. It combines shared channel patches with geometry-aware attention, restricts spectral targets to each recording's supported frequency band, and balances exposure across data sources. We pretrain a 109M-parameter model on 11 public EMG sources and evaluate it on gesture recognition, continuous-force regression, and contact classification. EMGBlend consistently outperforms matched random initialization and waveform reconstruction controls. Fixed-budget source controls show that multi-source pretraining improves gesture recognition and remains competitive for force decoding. Ablations confirm that geometry, band-aware targets, and source balancing each contribute to transfer, although cross-person NinaPro force estimation remains difficult. Overall, EMGBlend shows how heterogeneous EMG datasets can be combined through explicit mechanism design rather than simple concatenation. Code is available at https://github.com/tamanano/EMGBlend