EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Ensemble Network-Based Distillation for Hyperspectral Image Classification in the Presence of Label Noise

Youqiang Zhang, Ruihui Ding, Hao Shi, Jiaxi Liu, Qiqiong Yu, Guo Cao, Xuesong Li

Remote Sensing · 14.11.2024 · 16 (22) · 4247

DOI: 10.3390/rs16224247 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Remote Sensing, online erschienen am 14.11.2024. Verfasst von Youqiang Zhang und 6 weiteren Autor:innen. Die Arbeit wurde bisher 6-mal zitiert (Crossref) und nennt 84 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: National Natural Science Foundation of China, Open Project of Tianjin Key Laboratory of Autonomous Intelligence Technology and Systems, Natural Science Foundation of Jiangsu Province.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Deep learning has made remarkable strides in hyperspectral image (HSI) classification, significantly improving classification performance. However, the challenge of obtaining accurately labeled training samples persists, primarily due to the subjectivity of human annotators and their limited domain knowledge. This often results in erroneous labels, commonly referred to as label noise. Such noisy labels can critically impair the performance of deep learning models, making it essential to address this issue. While previous studies focused on label noise filtering and label correction, these approaches often require estimating noise rates and may inadvertently propagate noisy labels to clean labels, especially in scenarios with high noise levels. In this study, we introduce an ensemble network-based distillation (END) method specifically designed to address the challenges posed by label noise in HSI classification. The core idea is to leverage multiple base neural networks to generate an estimated label distribution from the training data. This estimated distribution is then used alongside the ground-truth labels to train the target network effectively. Moreover, we propose a parameter-adaptive loss function that balances the impact of both the estimated and ground-truth label distributions during the training process. Our approach not only simplifies architectural requirements but also integrates seamlessly into existing deep learning frameworks. Comparative experiments on four hyperspectral datasets demonstrate the effectiveness of our method, highlighting its competitive performance in the presence of label noise.

Förderung

  • National Natural Science Foundation of China
  • Open Project of Tianjin Key Laboratory of Autonomous Intelligence Technology and Systems
  • Natural Science Foundation of Jiangsu Province
  • Basic Research Project of the Natural Science Foundation of the Jiangsu Higher Education Institutions

Zitieren

APA

Zhang, Y., Ding, R., Shi, H., Liu, J., Yu, Q., Cao, G., & Li, X. (2024). Ensemble Network-Based Distillation for Hyperspectral Image Classification in the Presence of Label Noise. Remote Sensing, 16(22), 4247. https://doi.org/10.3390/rs16224247

Vancouver

Zhang Y, Ding R, Shi H, Liu J, Yu Q, Cao G, et al. Ensemble Network-Based Distillation for Hyperspectral Image Classification in the Presence of Label Noise. Remote Sensing. 2024;16(22):4247. doi:10.3390/rs16224247

BibTeX
@article{Zhang2024,
  author = {Zhang, Youqiang and Ding, Ruihui and Shi, Hao and Liu, Jiaxi and Yu, Qiqiong and Cao, Guo and Li, Xuesong},
  title = {Ensemble Network-Based Distillation for Hyperspectral Image Classification in the Presence of Label Noise},
  journal = {Remote Sensing},
  year = {2024},
  volume = {16},
  number = {22},
  pages = {4247},
  doi = {10.3390/rs16224247}
}
Kostenlos starten →