Scaling and Distilling Transformer Models for sEMG
This enables efficient and expressive models for complex real-time sEMG tasks in human-computer interfaces, though it is incremental as it applies existing scaling and distillation techniques to sEMG data.
The paper tackled the problem of limited training data and computational constraints in surface electromyography (sEMG) tasks by scaling transformer models up to 110M parameters, achieving improved cross-user performance, and distilling them into models 50x smaller with less than 1.5% absolute performance loss.
Surface electromyography (sEMG) signals offer a promising avenue for developing innovative human-computer interfaces by providing insights into muscular activity. However, the limited volume of training data and computational constraints during deployment have restricted the investigation of scaling up the model size for solving sEMG tasks. In this paper, we demonstrate that vanilla transformer models can be effectively scaled up on sEMG data and yield improved cross-user performance up to 110M parameters, surpassing the model size regime investigated in other sEMG research (usually <10M parameters). We show that >100M-parameter models can be effectively distilled into models 50x smaller with minimal loss of performance (<1.5% absolute). This results in efficient and expressive models suitable for complex real-time sEMG tasks in real-world environments.