Journal of Korean Society of Industrial and Systems Engineering (산업경영시스템학회지)

Society of Korea Industrial and System Engineering (한국산업경영시스템학회)
권호 표지
DOI QR코드

DOI QR Code

Performance Analysis of MixMatch-Based Semi-Supervised Learning for Defect Detection in Manufacturing Processes

제조 공정 결함 탐지를 위한 MixMatch 기반 준지도학습 성능 분석

  • Ye-Jun Kim (Department of Industrial Management Engineering, Kyonggi University) ;
  • Ye-Eun Jeong (Department of Industrial and Systems Engineering, Kyonggi University Graduate School) ;
  • Yong Soo Kim (Department of Industrial Management Engineering, Kyonggi University)
  • 김예준 (경기대학교 산업경영공학과) ;
  • 정예은 (경기대학교 일반대학원 산업시스템공학과) ;
  • 김용수 (경기대학교 산업경영공학과)
  • Received : 2023.10.28
  • Accepted : 2023.12.14
  • Published : 2023.12.31
Download PDF
⟨ Previous Next ⟩

Abstract

Recently, there has been an increasing attempt to replace defect detection inspections in the manufacturing industry using deep learning techniques. However, obtaining substantial high-quality labeled data to enhance the performance of deep learning models entails economic and temporal constraints. As a solution for this problem, semi-supervised learning, using a limited amount of labeled data, has been gaining traction. This study assesses the effectiveness of semi-supervised learning in the defect detection process of manufacturing using the MixMatch algorithm. The MixMatch algorithm incorporates three dominant paradigms in the semi-supervised field: Consistency regularization, Entropy minimization, and Generic regularization. The performance of semi-supervised learning based on the MixMatch algorithm was compared with that of supervised learning using defect image data from the metal casting process. For the experiments, the ratio of labeled data was adjusted to 5%, 10%, 25%, and 50% of the total data. At a labeled data ratio of 5%, semi-supervised learning achieved a classification accuracy of 90.19%, outperforming supervised learning by approximately 22%p. At a 10% ratio, it surpassed supervised learning by around 8%p, achieving a 92.89% accuracy. These results demonstrate that semi-supervised learning can achieve significant outcomes even with a very limited amount of labeled data, suggesting its invaluable application in real-world research and industrial settings where labeled data is limited.

Keywords

Acknowledgement

This work was supported by Kyonggi University's Graduate Research Assistantship 2023.

References

  1. Artificial Intelligence Index Report 2021, McKinsey & Company, Stanford University, 2021
  2. Berthelot, D., Carlini, N., Goodfellow, I., Papernot, N., Oliver, A., and Raffel, C.A., Mixmatch: A Holistic Approach to Semi-supervised Learning, Advances in Neural Information Processing Systems, 2019, Vol. 545, pp. 5049-5059.
  3. Casting Product Image Data for Quality Inspection, https://www.kaggle.com/ravirajsinh45/real-life-industrial-dataset-of-casting-product.
  4. Chun, C. and Ryu, S.K., Road Surface Damage Detection Based on Semi-supervised Learning Using Pseudo Labels, The Journal of the Korea Institute of Intelligent Transportation Systems, 2019, Vol. 18, No. 4, pp. 71-79. https://doi.org/10.12815/kits.2019.18.4.71
  5. Gao, D.,, Lin, X., Zheng, X., and Yang, Q., Novel Semi-supervised Fault Diagnosis Method Combining Tri-training and Deep Belief Network for Charging Equipment of Electric Vehicle, International Journal of Automotive Technology, 2022, Vol. 23, No. 6, pp. 1727-1737. https://doi.org/10.1007/s12239-022-0150-7
  6. Guo, C., Pleiss, G., Sun, Y., and Weinberger, K. Q., On calibration of modern neural networks, In International Conference on Machine Learning, PMLR, 2017, pp. 1321-1330.
  7. Grandvalet, Y. and Bengio, Y., Semi-supervised Learning by Entropy Minimization, Advances in Neural Information Processing Systems, 2004, Vol. 17, pp. 529-536.
  8. Kim, I.K., Kim, B.J., and Gwak, J.W., Utilizing Mean Teacher Semi-Supervised Learning for Robust Pothole Image Classification, Journal of The Korea Society of Computer and Information, 2023, Vol. 28, No. 5, pp. 17-27.
  9. Kingma, D.P., Mohamed, S., Jimenez Rezende, D., and Welling, M., Semi-supervised Learning with Deep Generative Models, Advances in Neural Information Processing Systems, 2014, Vol. 268, pp. 3581-3589.
  10. Kostopoulos, G., Karlos, S., Kotsiantis, S., and Ragos, O., Semi-supervised Regression: A Recent Review, Journal of Intelligent & Fuzzy Systems, 2018, Vol. 35, No. 2, pp. 1483-1500. https://doi.org/10.3233/JIFS-169689
  11. Laine, S. and Aila, T., Temporal Ensembling for Semi-Supervised Learning, arXiv preprint arXiv, 2016, 1610.02242.
  12. Lee, H.R., Hong, S.W., Lee, S.Y., and Ha, J.C., Anomaly Detection in Vehicle Network Using Semi-Supervised Learning Model, Journal of the Korea Academia-Industrial Cooperation Society, 2023, Vol. 24, No. 2, pp. 11-20. https://doi.org/10.5762/KAIS.2023.24.2.11
  13. Lee, S.W. and Kwon, Y.M., Machine Vision-based Quality Inspection Method and System Utilizing Deep Learning in Manufacturing Process, 10-2017-0162907, 2019.
  14. Lee, Y.H., Choi, J.E., and Hong, S.J., Semi-Supervised Learning for Fault Detection and Classification of Plasma Etch Equipment, Journal of the Semiconductor & Display Technology, 2020, Vol. 19, No. 4, pp. 121-125.
  15. Quali, Y., Hudelot, C., and Tami, M., An Overview of Deep Semi-supervised Learning, arXiv preprint arXiv, 2020, 2006.05278.
  16. Shim, S.B., Choi, S.I., Kong, S.M., and Lee, S.W., Adversarial Learning for Underground Structure Concrete Crack Detection Based on Semi-supervised Semantic Segmentation, Journal of Korean Tunnelling and Underground Space Association, Vol. 25, No. 2, pp. 515-528.
  17. Shin, Y.S., Kim, I.M., Kang, S.H., Lee, S.H., Park, J.H., and Kang, J.M, AMFM: AugMix-empowered FixMatch for High-Performing Semi-Supervised Classification of Metal Surface Defects, Journal of Korean Institute of Intelligent Systems, 2022, Vol. 32, No. 5, pp. 424-431. https://doi.org/10.5391/JKIIS.2022.32.5.424
  18. Sun, S., Zhou, Y., He, Hao., He, J., and Chi, Y., Anomaly Detection of Electricity Load Data Based on MixMatch, International Conference on Computer Engineering and Artificial Intelligence, Shanghai, China, 2021, pp. 53-56.
  19. Yang, X., Song, Z., King, I., and Xu, Z., A Survey on Deep Semi-supervised Learning, IEEE Transactions on Knowledge and Data Engineering, 2022, Vol. 35, No. 9, pp. 8934-8954. https://doi.org/10.1109/TKDE.2022.3220219
  20. Zagoruyko, S. and Komodakis, N., Wide Residual Networks, arXiv preprint arXiv, 2016, 1605.07146.
  21. Zhang, H., Cisse, M., Dauphin, Y.N., and Lopez-Paz, D., Mixup: Beyond Empirical Risk Minimization, arXiv preprint arXiv, 2017, 1710.09412.