Conclusion
Addressing the classification performance challenge in Φ-OTDR real-world applications due to the difficulty in obtaining enough labeled samples, we introduced and researched semi-supervised learning models tailored for Φ-OTDR event classification. Specifically, the XM-based models exhibit notable improvements in classification performance compared with the ST model based on pseudolabeling and the MT model based on consistency regularization. The EM approach introduces the SAT mechanism and incorporates SAF to encourage the model to make more accurate predictions for each class. This strategy generates meaningful adaptive thresholds, leading to further performance improvement over the XM. The proposed semi-supervised methods in this study offer the potential to enhance the accuracy of disturbance event classification in environmental settings without increasing optical equipment costs or complexity. These methods exhibit fast convergence and ease of transfer ability. The rapid convergence, ease of transferability, and significant performance enhancements exhibited by these methods position them as promising technological pathways for advancing the field of Φ-OTDR event classification, suggesting their potential to play a crucial role in future research and applications.
This is a preview of subscription content, log in via an institution to check access.
References
Grollmisch S, Cano E. Improving semi-supervised learning for audio classification with FixMatch. Electronics, 2021, 10: 1807
Wang Y D, Chen H, Heng Q, et al. FreeMatch: self-adaptive thresholding for semi-supervised learning. In: Proceedings of the International Conference on Learning Representation, Kigali, 2023
Sohn K, Berthelot D, Carlini N, et al. FixMatch: simplifying semi-supervised learning with consistency and confidence. In: Proceedings of the International Conference on Advances in Neural Information Processing Systems, 2020. 33: 596–608
Li Y J, Cao X M, Ni W H, et al. A deep learning model enabled multi-event recognition for distributed optical fiber sensing. Sci China Inf Sci, 2024, 67: 132404
Lee D H. Pseudo-label: the simple and efficient semi-supervised learning method for deep neural networks. In: Proceedings of Workshop on Challenges in Representation Learning, Atlanta, 2013. 3: 896
Acknowledgements
This work was supported in part by National Key Research and Development Program of China (Grant No. 2021YFB2900704) and Fundamental Research Funds for Central Universities (Grant No. 021314380211).
Author information
Authors and Affiliations
Corresponding author
Additional information
Supporting information Appendixes A–C. The supporting information is available online at info.scichina.com and link.springer.com. The supporting materials are published as submitted, without typesetting or editing. The responsibility for scientific accuracy and content remains entirely with the authors.
Rights and permissions
About this article
Cite this article
Li, Y., Hu, L. & Yu, K. Unleashing potentials with deep learning: decoding the complex events for distributed fiber optic sensing applications. Sci. China Inf. Sci. 67, 159402 (2024). https://doi.org/10.1007/s11432-023-3985-y
Received:
Revised:
Accepted:
Published:
DOI: https://doi.org/10.1007/s11432-023-3985-y