In Shift and In Variance: Assessing the Robustness of HAR Deep Learning Models Against Variability
Sensors · 2025
DOI: 10.3390/s25020430 ↗Kurz gefasst
Fachartikel in der Zeitschrift Sensors, 2025. Verfasst von Azhar Ali Khaked und 3 weiteren Autor:innen. Beteiligte Einrichtung: Department of Electrical and Computer Engineering, Concordia University, Montreal, QC H3G 1M8, Canada. Hauptthemen laut MEDLINE: Deep Learning, Human Activities. Die Arbeit wurde bisher 11-mal zitiert (Crossref) und nennt 43 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Concordia Startup.
Abstract (Originalsprache)
Deep learning (DL)-based Human Activity Recognition (HAR) using wearable inertial measurement unit (IMU) sensors can revolutionize continuous health monitoring and early disease prediction. However, most DL HAR models are untested in their robustness to real-world variability, as they are trained on limited lab-controlled data. In this study, we isolated and analyzed the effects of the subject, device, position, and orientation variabilities on DL HAR models using the HARVAR and REALDISP datasets. The Maximum Mean Discrepancy (MMD) was used to quantify shifts in the data distribution caused by these variabilities, and the relationship between the distribution shifts and model performance was drawn. Our HARVAR results show that different types of variability significantly degraded the DL model performance, with an inverse relationship between the data distribution shifts and performance. The compounding effect of multiple variabilities studied using REALDISP further underscores the challenges of generalizing DL HAR models to real-world conditions. Analyzing these impacts highlights the need for more robust models that generalize effectively to real-world settings. The MMD proved valuable for explaining the performance drops, emphasizing its utility in evaluating distribution shifts in HAR data.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Stichwörter der Autor:innen
Autor:innen und Institutionen
- Department of Electrical and Computer Engineering, Concordia University, Montreal, QC H3G 1M8, Canada.
- School of Engineering and Informatics, University of Sussex, Brighton BN1 9PS, UK.
ORCID-Profile: Azhar Ali Khaked, Nobuyuki Oishi, Daniel Roggen, Paula Lago
Förderung
- Concordia Startup
Zitieren
APA
Khaked, A. A., Oishi, N., Roggen, D., & Lago, P. (2025). In Shift and In Variance: Assessing the Robustness of HAR Deep Learning Models Against Variability. Sensors. https://doi.org/10.3390/s25020430
Vancouver
Khaked AA, Oishi N, Roggen D, Lago P. In Shift and In Variance: Assessing the Robustness of HAR Deep Learning Models Against Variability. Sensors. 2025. doi:10.3390/s25020430
BibTeX
@article{Khaked2025,
author = {Khaked, Azhar Ali and Oishi, Nobuyuki and Roggen, Daniel and Lago, Paula},
title = {In Shift and In Variance: Assessing the Robustness of HAR Deep Learning Models Against Variability},
journal = {Sensors},
year = {2025},
doi = {10.3390/s25020430}
}
EUVIMED