CLF-AIAD : a contrastive learning framework for acoustic industrial anomaly detection
- Author
- Zhaoyi Liu, Yuanbo Hou (UGent) , Haoyu Tang, Álvaro López-Chilet, Sam Michiels, Dick Botteldooren (UGent) , Jon Ander Gómez and Danny Hughes
- Organization
- Abstract
- Acoustic Industrial Anomaly Detection (AIAD) has received a great deal of attention as a technique to discover faults or malicious activity, allowing for preventive measures to be more effectively targeted. The essence of AIAD is to learn the compact distribution of normal acoustic data and detect outliers as anomalies during testing. However, recent AIAD work does not capture the dependencies and dynamics of Acoustic Industrial Data (AID). To address this issue, we propose a novel Contrastive Learning Framework (CLF) for AIAD, known as CLF-AIAD. Our method introduces a multi-grained contrastive learning-based framework to extract robust normal AID representations. Specifically, we first employ a projection layer and a novel context-based contrast method to learn robust temporal vectors. Building upon this, we then introduce a sample-wise contrasting-based module to capture local invariant characteristics, improving the discriminative capabilities of the model. Finally, a transformation classifier is introduced to bolster the performance of the primary task under a self-supervised learning framework. Extensive experiments on two typical industrial datasets, MIMII and ToyADMOS, demonstrate that our proposed CLF-AIAD effectively detects various real-world defects and improves upon the state-of-the-art in unsupervised industrial anomaly detection.
Downloads
-
(...).pdf
- full text (Published version)
- |
- UGent only
- |
- |
- 1.78 MB
-
(...).pdf
- full text (Accepted manuscript)
- |
- UGent only (changes to open access on 2024-11-14)
- |
- |
- 2.32 MB
Citation
Please use this url to cite or link to this publication: http://hdl.handle.net/1854/LU-01HP21FCWETYV4SYSQWM6RK4K7
- MLA
- Liu, Zhaoyi, et al. “CLF-AIAD : A Contrastive Learning Framework for Acoustic Industrial Anomaly Detection.” Neural Information Processing : 30th International Conference, ICONIP 2023, Proceedings, Part VII, edited by Biao Luo et al., vol. 1961, Springer, 2024, pp. 125–37, doi:10.1007/978-981-99-8126-7_10.
- APA
- Liu, Z., Hou, Y., Tang, H., López-Chilet, Á., Michiels, S., Botteldooren, D., … Hughes, D. (2024). CLF-AIAD : a contrastive learning framework for acoustic industrial anomaly detection. In B. Luo, L. Cheng, Z.-G. Wu, H. Li, & C. Li (Eds.), Neural Information Processing : 30th International Conference, ICONIP 2023, Proceedings, Part VII (Vol. 1961, pp. 125–137). https://doi.org/10.1007/978-981-99-8126-7_10
- Chicago author-date
- Liu, Zhaoyi, Yuanbo Hou, Haoyu Tang, Álvaro López-Chilet, Sam Michiels, Dick Botteldooren, Jon Ander Gómez, and Danny Hughes. 2024. “CLF-AIAD : A Contrastive Learning Framework for Acoustic Industrial Anomaly Detection.” In Neural Information Processing : 30th International Conference, ICONIP 2023, Proceedings, Part VII, edited by Biao Luo, Long Cheng, Zheng-Guang Wu, Hongyi Li, and Chaojie Li, 1961:125–37. Springer. https://doi.org/10.1007/978-981-99-8126-7_10.
- Chicago author-date (all authors)
- Liu, Zhaoyi, Yuanbo Hou, Haoyu Tang, Álvaro López-Chilet, Sam Michiels, Dick Botteldooren, Jon Ander Gómez, and Danny Hughes. 2024. “CLF-AIAD : A Contrastive Learning Framework for Acoustic Industrial Anomaly Detection.” In Neural Information Processing : 30th International Conference, ICONIP 2023, Proceedings, Part VII, ed by. Biao Luo, Long Cheng, Zheng-Guang Wu, Hongyi Li, and Chaojie Li, 1961:125–137. Springer. doi:10.1007/978-981-99-8126-7_10.
- Vancouver
- 1.Liu Z, Hou Y, Tang H, López-Chilet Á, Michiels S, Botteldooren D, et al. CLF-AIAD : a contrastive learning framework for acoustic industrial anomaly detection. In: Luo B, Cheng L, Wu Z-G, Li H, Li C, editors. Neural Information Processing : 30th International Conference, ICONIP 2023, Proceedings, Part VII. Springer; 2024. p. 125–37.
- IEEE
- [1]Z. Liu et al., “CLF-AIAD : a contrastive learning framework for acoustic industrial anomaly detection,” in Neural Information Processing : 30th International Conference, ICONIP 2023, Proceedings, Part VII, Changsha, China, 2024, vol. 1961, pp. 125–137.
@inproceedings{01HP21FCWETYV4SYSQWM6RK4K7,
abstract = {{Acoustic Industrial Anomaly Detection (AIAD) has received a great deal of attention as a technique to discover faults or malicious activity, allowing for preventive measures to be more effectively targeted. The essence of AIAD is to learn the compact distribution of normal acoustic data and detect outliers as anomalies during testing. However, recent AIAD work does not capture the dependencies and dynamics of Acoustic Industrial Data (AID). To address this issue, we propose a novel Contrastive Learning Framework (CLF) for AIAD, known as CLF-AIAD. Our method introduces a multi-grained contrastive learning-based framework to extract robust normal AID representations. Specifically, we first employ a projection layer and a novel context-based contrast method to learn robust temporal vectors. Building upon this, we then introduce a sample-wise contrasting-based module to capture local invariant characteristics, improving the discriminative capabilities of the model. Finally, a transformation classifier is introduced to bolster the performance of the primary task under a self-supervised learning framework. Extensive experiments on two typical industrial datasets, MIMII and ToyADMOS, demonstrate that our proposed CLF-AIAD effectively detects various real-world defects and improves upon the state-of-the-art in unsupervised industrial anomaly detection.}},
author = {{Liu, Zhaoyi and Hou, Yuanbo and Tang, Haoyu and López-Chilet, Álvaro and Michiels, Sam and Botteldooren, Dick and Gómez, Jon Ander and Hughes, Danny}},
booktitle = {{Neural Information Processing : 30th International Conference, ICONIP 2023, Proceedings, Part VII}},
editor = {{Luo, Biao and Cheng, Long and Wu, Zheng-Guang and Li, Hongyi and Li, Chaojie}},
isbn = {{9789819981250}},
issn = {{1865-0929}},
language = {{eng}},
location = {{Changsha, China}},
pages = {{125--137}},
publisher = {{Springer}},
title = {{CLF-AIAD : a contrastive learning framework for acoustic industrial anomaly detection}},
url = {{http://doi.org/10.1007/978-981-99-8126-7_10}},
volume = {{1961}},
year = {{2024}},
}
- Altmetric
- View in Altmetric