EUVIMEDEUVIMED entdecken
Crossref

Hierarchical Unsupervised Feature Extraction and Fusion of Multisensor for Human Activity Recognition

Saeid Arabzadeh, Farshad Almasganj, Mohammad Mahdi Ahmadi

IEEE Sensors Journal · 15.05.2026 · 26 (10) · 15595-15607

DOI: 10.1109/jsen.2026.3668934 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Sensors Journal, erschienen am 15.05.2026. Verfasst von Saeid Arabzadeh und 2 weiteren Autor:innen. Die Arbeit wurde bisher 2-mal zitiert (Crossref).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Deep learning methods are widely used for human activity recognition (HAR) with inertial measurement unit (IMU) sensors placed on various parts of the human body. In such systems, analyzing data from multiple IMU sensors presents a significant challenge. Fusing outputs from individual sensors, as well as integrating information across sensors, is a complex task that often poses challenges for simple one-stage models, particularly in explicitly capturing distinct intrasensor and intersensor. In this work, we propose a comprehensive hierarchical unsupervised fusion (HUF) method for feature extraction and fusion from IMU sensor data, using a hybrid architecture of convolutional neural networks (CNNs) and autoencoders (AEs). First, a stacked CNN-AE encodes short-time signals into high-dimensional feature representations. Second, a dedicated CNN-AE performs intrasensory fusion by combining features within each sensor. Finally, a third CNN-AE—preceded by a custom-designed attention mechanism (AM)—fuses features across all sensors for intersensory fusion. The hierarchical structure enables the AM to calibrate sensor contributions based on activity type, thereby enhancing the fusion process. The proposed HUF method achieves 96.8% accuracy on the UCI-HAR dataset—2.8% higher than state-of-the-art supervised methods. It also achieves 92.9% and 87.6% accuracy on the daily life activity (DaLiAc) and Parkinson’s disease (PD) gait multimodal dataset of freezing of gait in PD (MDFG-PD) datasets, respectively. These results further improve to 94.8% and 89.0% when the AM is integrated into the HUF framework.

Zitieren

APA

Arabzadeh, S., Almasganj, F., & Ahmadi, M. M. (2026). Hierarchical Unsupervised Feature Extraction and Fusion of Multisensor for Human Activity Recognition. IEEE Sensors Journal, 26(10), 15595-15607. https://doi.org/10.1109/jsen.2026.3668934

Vancouver

Arabzadeh S, Almasganj F, Ahmadi MM. Hierarchical Unsupervised Feature Extraction and Fusion of Multisensor for Human Activity Recognition. IEEE Sensors Journal. 2026;26(10):15595-15607. doi:10.1109/jsen.2026.3668934

BibTeX
@article{Arabzadeh2026,
  author = {Arabzadeh, Saeid and Almasganj, Farshad and Ahmadi, Mohammad Mahdi},
  title = {Hierarchical Unsupervised Feature Extraction and Fusion of Multisensor for Human Activity Recognition},
  journal = {IEEE Sensors Journal},
  year = {2026},
  volume = {26},
  number = {10},
  pages = {15595--15607},
  doi = {10.1109/jsen.2026.3668934}
}
Kostenlos starten →