Detecting Domain Shifts in Myoelectric Activations: Challenges and Opportunities in Stream Learning

2025-08-28

AI Summary

To address domain shift detection challenges arising from the non-stationarity of electromyographic (EMG) signals, this paper proposes a real-time decoding framework based on data stream learning. Methodologically, we introduce a cosine-kernel kernel principal component analysis (KPCA) preprocessing strategy to enhance discriminative representation of domain discrepancies; systematically benchmark mainstream drift detection algorithms—including CUSUM, Page-Hinckley, and ADWIN—on the Ninapro DB6 dataset; and incorporate a temporal segmentation modeling mechanism to capture dynamic cross-subject and cross-session domain evolution. Results reveal that existing drift detectors exhibit low sensitivity and high false-positive rates in EMG streaming scenarios. Beyond identifying these performance bottlenecks, our work validates the efficacy of an integrated stream learning paradigm—combining kernel-based feature learning with adaptive drift detection—for sustaining long-term decoding stability. This advances robust online myoelectric interfacing.

Technology Category

• Data Mining & Knowledge Management: Data Stream Mining
• Machine Learning: Kernel Methods
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Data management and stream processing for Web, mobile and wireless applications
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods

Abstract

Detecting domain shifts in myoelectric activations poses a significant challenge due to the inherent non-stationarity of electromyography (EMG) signals. This paper explores the detection of domain shifts using data stream (DS) learning techniques, focusing on the DB6 dataset from the Ninapro database. We define domains as distinct time-series segments based on different subjects and recording sessions, applying Kernel Principal Component Analysis (KPCA) with a cosine kernel to pre-process and highlight these shifts. By evaluating multiple drift detection methods such as CUSUM, Page-Hinckley, and ADWIN, we reveal the limitations of current techniques in achieving high performance for real-time domain shift detection in EMG signals. Our results underscore the potential of streaming-based approaches for maintaining stable EMG decoding models, while highlighting areas for further research to enhance robustness and accuracy in real-world scenarios.

Problem

• Detecting domain shifts in myoelectric EMG signals due to non-stationarity
• Evaluating real-time drift detection methods for EMG signal stability
• Enhancing robustness of streaming-based EMG decoding models in practice

Innovation

• KPCA with cosine kernel for preprocessing
• Evaluating multiple drift detection methods
• Stream learning for EMG domain shifts