EUVIMEDEUVIMED entdecken
Crossref

Hierarchical Transformer With Auxiliary Learning for Subject-Independent Respiration Emotion Recognition

Yong Wang, Chendong Xu, Weirui Na, Dongyu Liu, Jiuqi Yan, Shuai Yao, Qisong Wu

IEEE Sensors Journal · 15.08.2025 · 25 (16) · 31290-31301

DOI: 10.1109/jsen.2025.3587271 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Sensors Journal, erschienen am 15.08.2025. Verfasst von Yong Wang und 6 weiteren Autor:innen. Die Arbeit wurde bisher 10-mal zitiert (Crossref) und nennt 60 Literaturangaben. Gefördert von: National Natural Science Foundation of China, Key Laboratory of Underwater Acoustic Countermeasure Technology.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Respiration is modulated by human emotional activity. Emotion recognition using physiological signals has recently gained considerable attention. However, most existing studies primarily focus on using electroencephalogram (EEG) signals for emotion recognition. This paper explores the potential of utilizing respiration signals collected by wearable devices for emotion recognition. We propose a hierarchical Transformer model to effectively extract emotional information from respiration signals. Furthermore, we introduce gender classification as an auxiliary task to further improve the accuracy of emotion recognition. Specifically, a Frame Transformer is employed to capture emotional information across frames of respiration signals. The extracted frame-level features are subsequently fused with segment-level embeddings through a specially designed fusion layer. Next, separate Segment Transformers are employed for emotion and gender to extract segment-level information, with a cosine similarity loss applied to promote shared feature learning. Finally, distinct classifiers are used for emotion and gender classification. To evaluate the effectiveness of the proposed method, extensive experiments are conducted on the DEAP and MAHNOB-HCI datasets under a subject-independent setting. For the DEAP dataset, the average classification accuracies are 72.42% for valence and 73.91% for arousal, while for the MAHNOB-HCI dataset, they are 80.45% and 79.69%, respectively. Compared to other physiological signals such as EEG, respiration signals exhibit comparable potential for emotion recognition.

Förderung

  • National Natural Science Foundation of China
  • Key Laboratory of Underwater Acoustic Countermeasure Technology

Zitieren

APA

Wang, Y., Xu, C., Na, W., Liu, D., Yan, J., Yao, S., & Wu, Q. (2025). Hierarchical Transformer With Auxiliary Learning for Subject-Independent Respiration Emotion Recognition. IEEE Sensors Journal, 25(16), 31290-31301. https://doi.org/10.1109/jsen.2025.3587271

Vancouver

Wang Y, Xu C, Na W, Liu D, Yan J, Yao S, et al. Hierarchical Transformer With Auxiliary Learning for Subject-Independent Respiration Emotion Recognition. IEEE Sensors Journal. 2025;25(16):31290-31301. doi:10.1109/jsen.2025.3587271

BibTeX
@article{Wang2025,
  author = {Wang, Yong and Xu, Chendong and Na, Weirui and Liu, Dongyu and Yan, Jiuqi and Yao, Shuai and Wu, Qisong},
  title = {Hierarchical Transformer With Auxiliary Learning for Subject-Independent Respiration Emotion Recognition},
  journal = {IEEE Sensors Journal},
  year = {2025},
  volume = {25},
  number = {16},
  pages = {31290--31301},
  doi = {10.1109/jsen.2025.3587271}
}
Kostenlos starten →