Analysis of Latent Defect Detection Using Sigma Deviation Count Labeling (SDCL)
Abstract
1. Introduction
2. Related Work
2.1. Outlier Detection Based on the Normal Distribution
2.2. Mean Absolute Deviation
2.3. The Interquartile Range
3. Data Set
4. Data Preprocessing
4.1. Data Cleaning
4.2. Outiler Count and Labeling
5. Performance Results
6. Conclusions
7. Discussion
Author Contributions
Funding
Data Availability Statement
Acknowledgments
Conflicts of Interest
References
- Nguyen, M.H.; Kwak, S. Enhance Reliability of Semiconductor Devices in Power Converters. Electronics 2020, 9, 2068. [Google Scholar] [CrossRef]
- Liao, H.T. Reliability prediction and testing plan based on an accelerated degradation testing model. Int. J. Mater. Prod. Technol. Technol. 2004, 21, 402–422. [Google Scholar] [CrossRef]
- Julitz, T.M.; Schlüter, N.; Löwer, M. Scenario-Based Failure Analysis of Product Systems and their Environment. arXiv 2023, arXiv:2306.15694. [Google Scholar] [CrossRef]
- Kim, J.; Joe, I. Chip-Level Defect Analysis with Virtual Bad Wafers Based on Huge Big Data Handling for Semiconductor Production. Electronics 2024, 13, 2205. [Google Scholar] [CrossRef]
- Price, D.W.; Sutherland, D.G.; Rathert, J. Process Watch: The (Automotive) Problem with Semiconductors. Solid State Technol. 2018, 15, 1–5. [Google Scholar]
- Qiao, X.; Jauw, V.L.; Seong, L.C.; Banda, T. Advances and limitations in machine learning approaches applied to remaining useful life predictions: A critical review. Int. J. Adv. Manuf. Technol. 2024, 133, 4059–4076. [Google Scholar] [CrossRef]
- Ooi, M.P.-L.; Kassim, Z.A.; Demidenko, S. Shortening Burn-in Test: Application of Weibull Statistical Analysis & HVST. In Proceedings of the 2005 IEEE Instrumentation and Measurement Technology Conference, Ottawa, ON, Canada, 16–19 May 2005; Volume 1, pp. 1–6. [Google Scholar] [CrossRef]
- Suhir, E. To Burn-In, or Not to Burn-In: That’s the Question. Aerospace 2019, 6, 29. [Google Scholar] [CrossRef]
- Wang, M. A Review of Reliability in Gate-All-Around Nanosheet Devices. Micromachines 2024, 15, 269. [Google Scholar] [CrossRef]
- Lenhard, P.; Kovalenko, A.; Lenhard, R. Die Level Predictive Modeling to Reduce Latent Reliability Defect Escapes. Microelectron. Reliab. 2023, 148, 115139. [Google Scholar] [CrossRef]
- Moreno-Lizaranzu, M.J.; Cuesta, F. Improving Electronic Sensor Reliability by Robust Outlier Screening. Sensors 2013, 13, 13521–13542. [Google Scholar] [CrossRef] [PubMed]
- Pihlaja, D. Real Time Dynamic Application of Part Average Testing (PAT) at Final Test. In Proceedings of the CS MANTECH Conference, New Orleans, LA, USA, 13–16 May 2013; pp. 165–167. [Google Scholar]
- Wang, C.-C.; Yang, Y.-Y. A Machine Learning Approach for Improving Wafer Acceptance Testing Based on an Analysis of Station and Equipment Combinations. Mathematics 2023, 11, 1569. [Google Scholar] [CrossRef]
- Hu, H.; Patel, S.; Hsiao, H.; Tretz, F.; Volk, T.; Arslan, M.; Nix, R.; Jindal, S.; Archambeault, B. Advanced Outlier Detection Using Unsupervised Learning for Screening Potential Customer Returns. In Proceedings of the 2020 IEEE International Test Conference, Washington, DC, USA, 2–5 November 2020; pp. 1–10. [Google Scholar] [CrossRef]
- van Selst, M.; Jolicoeur, P. A Solution to the Effect of Sample Size on Outlier Elimination. Q. J. Exp. Psychol. Sect. 1994, 50, 386–393. [Google Scholar] [CrossRef]
- Yang, J.; Rahardja, S.; Fränti, P. Outlier Detection: How to Threshold Outlier Scores? In Proceedings of the International Conference on Artificial Intelligence, Information Processing and Cloud Computing, New York, NY, USA, 19–21 December 2019; pp. 1–6. [Google Scholar]
- Wan, X.; Wang, W.; Liu, J.; Tong, T. Estimating the Sample Mean and Standard Deviation from the Sample Size, Median, Range and/or Interquartile Range. BMC Med. Res. Methodol. 2014, 14, 135. [Google Scholar] [CrossRef]
- Dhwani, D.; Varma, T. A Review of Various Statistical Methods for Outlier Detection. Int. J. Comput. Sci. Eng. Technol. 2014, 5, 137–140. [Google Scholar]
- Wickham, H.; Stryjewski, L. 40 Years of Boxplots. Am. Stat. 2011, 65, 1–6. [Google Scholar]
- Jones, P.R. A Note on Detecting Statistical Outliers in Psychophysical Data. Atten. Percept. Psychophys. 2019, 81, 1189–1196. [Google Scholar] [CrossRef] [PubMed]
- Atif, M.; Farooq, M.; Shafiq, M.; Alballa, T.; Alhabeeb, S.A.; Khalifa, H.A.-E.-W. Uncovering the Impact of Outliers on Clusters’ Evolution in Temporal Data-Sets: An Empirical Analysis. Sci. Rep. 2024, 14, 30674. [Google Scholar] [CrossRef]
- França, C.M.; Couto, R.S.; Velloso, P.B. Missing Data Imputation in Internet of Things Gateways. Information 2021, 12, 425. [Google Scholar] [CrossRef]
- Emmanuel, T.; Tlamelo, M.; Phaneendra, B.; Das, D.; Epule, E. A Survey on Missing Data in Machine Learning. J. Big Data 2021, 8, 140. [Google Scholar] [CrossRef]
- Kang, H. The Prevention and Handling of the Missing Data. Korean J. Anesth. 2013, 64, 402–406. [Google Scholar] [CrossRef]
- Chandrashekar, G.; Sahin, F. A Survey on Feature Selection Methods. Comput. Electr. Eng. 2014, 40, 16–28. [Google Scholar] [CrossRef]
- Jizat, J.A.M.; Ahmad, N.; Alwi, N.H.; Nor, M.I.; Salleh, N.M.; Kamal, M.M. Evaluation of the Machine Learning Classifier in Wafer Defects Classification. ICT Express 2021, 7, 535–539. [Google Scholar] [CrossRef]
- Hu, J.; Zhou, Z.; Chen, J.; Wang, Z.; Lin, S.; Li, Y. A Novel Quality Prediction Method Based on Feature Selection Considering High Dimensional Product Quality Data. J. Ind. Manag. Optim. 2022, 18, 2715–2735. [Google Scholar] [CrossRef]
- Taha, K. Observational and Experimental Insights into Machine Learning-Based Defect Classification in Wafers. J. Intell. Manuf. 2025, 020502. [Google Scholar] [CrossRef]
- Lee, Y.; Roh, Y. An Expandable Yield Prediction Framework Using Explainable Artificial Intelligence for Semiconductor Manufacturing. Appl. Sci. 2023, 13, 2660. [Google Scholar] [CrossRef]
- Barandela, R.; Valdovinos, R.M.; Sánchez, J.S.; Ferri, F.J. The Imbalanced Training Sample Problem: Under or Over Sampling? In Proceedings of the Structural, Syntactic, and Statistical Pattern Recognition, Lisbon, Portugal, 18–20 August 2004; Structural, Syntactic, and Statistical Pattern Recognition. Springer: Berlin/Heidelberg, Germany, 2004; Volume 3138, pp. 806–814. [Google Scholar]
- García, V.; Sánchez, J.S.; Mollineda, R.A. On the Effectiveness of Preprocessing Methods When Dealing with Different Levels of Class Imbalance. Knowl.-Based Syst. 2012, 25, 13–21. [Google Scholar] [CrossRef]
Range | Probability Within the Range | Probability Outside the Range |
---|---|---|
μ ± 1σ | 68.27% | 33.73% |
μ ± 2σ | 95.45% | 4.55% |
μ ± 3σ | 99.73% | 0.27% |
Dataset | Total Samples | Features (Before Cleaning) | Features (After Cleaning) |
---|---|---|---|
1st Dataset | 14,140 | 1804 | 1664 |
2nd Dataset | 80,819 | 454 | 374 |
Process | Dataset | Counting | Threshold | Label 0 | Label 1 |
---|---|---|---|---|---|
1 | 1st Dataset | MAD | 3 | 12,453 (88.09%) | 1687 (11.91%) |
2 | 2nd Dataset | MAD | 3 | 72,818 (90.10%) | 8001 (9.89%) |
3 | 1st Dataset | IQR | 1.5 | 13,103 (92.67%) | 1037 (7.33%) |
4 | 2nd Dataset | IQR | 1.5 | 76,866 (95.10%) | 3953 (4.89%) |
5 | 1st Dataset | Sigma | 2σ | 13,490 (95.40%) | 650 (4.60%) |
6 | 2nd Dataset | Sigma | 2σ | 77,231 (95.56%) | 3588 (4.44%) |
7 | 1st Dataset | Sigma | 3σ | 13,045 (99.73%) | 35 (0.27%) |
8 | 2nd Dataset | Sigma | 3σ | 28,895 (99.74%) | 76 (0.26%) |
Process | Model | Scaler | Accuracy (%) | GM (%) 2 | Sensitivity (%) | Specificity (%) |
---|---|---|---|---|---|---|
Process 1 | XGB | Normalize | 97.5 | 86.4 | 75.5 | 98.8 |
Process 2 | MinMax | 96.3 | 84.9 | 72.8 | 98.9 | |
Process 1 | SVM | MinMax | 97.2 | 86.7 | 76.4 | 98.4 |
Process 2 | MinMax | 92.1 | 84.5 | 76.1 | 93.9 | |
Process 1 | LR | None | 91.2 | 88.6 | 85.8 | 91.5 |
Process 2 | Standard | 87.4 | 84.8 | 81.7 | 88.0 | |
Process 1 | KNN | None | 87.3 | 88.2 | 89.3 | 87.2 |
Process 2 | None | 76.4 | 75.5 | 74.4 | 76.6 | |
Process 1 | DT | None | 93.9 | 80.2 | 67.4 | 95.4 |
Process 2 | None | 90.8 | 78.8 | 66.3 | 93.5 | |
Process 1 | ADA | None | 94.4 | 84.3 | 74.2 | 95.6 |
Process 2 | None | 92.6 | 80.0 | 67.1 | 95.4 |
Process | Model | Scaler | Accuracy (%) | GM (%) 2 | Sensitivity (%) | Specificity (%) |
---|---|---|---|---|---|---|
Process 3 | XGB | MinMax | 96.7 | 82.3 | 68.5 | 98.9 |
Process 4 | MinMax | 98.4 | 88.4 | 78.5 | 99.5 | |
Process 3 | SVM | MinMax | 95.8 | 82.8 | 70.1 | 97.8 |
Process 4 | MinMax | 96.8 | 91.6 | 86.3 | 97.3 | |
Process 3 | LR | None | 88.3 | 85.0 | 81.4 | 88.9 |
Process 4 | MinMax | 91.3 | 88.5 | 85.5 | 91.6 | |
Process 3 | KNN | None | 81.8 | 82.0 | 82.3 | 81.8 |
Process 4 | MinMax | 81.4 | 87.8 | 95.5 | 80.7 | |
Process 3 | DT | None | 91.3 | 77.5 | 64.3 | 93.5 |
Process 4 | None | 96.3 | 83.8 | 71.9 | 97.6 | |
Process 3 | ADA | Normalize | 91.8 | 81.7 | 71.4 | 93.5 |
Process 4 | Normalize | 94.9 | 83.8 | 73.1 | 96.0 |
Process | Model | Scaler | Accuracy (%) | GM (%) 2 | Sensitivity (%) | Specificity (%) |
---|---|---|---|---|---|---|
Process 5 | XGB | MinMax | 99.0 | 94.5 | 89.8 | 99.4 |
Process 6 | None | 99.8 | 44.7 | 20.0 | 100 | |
Process 5 | SVM | MinMax | 98.5 | 92.7 | 86.7 | 99.1 |
Process 6 | None | 97.4 | 95.7 | 93.8 | 97.5 | |
Process 5 | LR | None | 92.5 | 90.5 | 88.3 | 92.7 |
Process 6 | Normalize | 87.8 | 72.6 | 60.0 | 87.8 | |
Process 5 | KNN | None | 96.3 | 95.5 | 94.5 | 96.4 |
Process 6 | None | 99.6 | 89.3 | 80.0 | 99.6 | |
Process 5 | DT | MinMax | 97.7 | 89.5 | 81.3 | 98.5 |
Process 6 | None | 99.7 | 44.7 | 20.0 | 99.8 | |
Process 5 | ADA | None | 98.6 | 94.3 | 89.8 | 99.0 |
Process 6 | Normalize | 100 | 89.4 | 80.0 | 100 |
Process | Model | Scaler | Accuracy (%) | GM (%) 2 | Sensitivity (%) | Specificity (%) |
---|---|---|---|---|---|---|
Process 7 | XGB | None | 98.7 | 91.9 | 85.0 | 99.3 |
Process 8 | None | 100 | 100 | 100 | 100 | |
Process 7 | SVM | Standard | 97.4 | 88.3 | 80.0 | 97.4 |
Process 8 | Standard | 99.3 | 99.6 | 100 | 99.2 | |
Process 7 | LR | Standard | 93.0 | 91.7 | 90.4 | 93.1 |
Process 8 | MinMax | 99.9 | 100 | 100 | 99.9 | |
Process 7 | KNN | None | 94.4 | 93.7 | 92.9 | 94.4 |
Process 8 | Standard | 99.8 | 99.9 | 100 | 99.8 | |
Process 7 | DT | Normalize | 97.2 | 86.4 | 76.0 | 98.3 |
Process 8 | Standard | 100 | 79.0 | 62.5 | 100 | |
Process 7 | ADA | MinMax | 98.3 | 94.3 | 90.1 | 98.7 |
Process 8 | None | 100 | 93.5 | 87.5 | 100 |
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. |
© 2025 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).
Share and Cite
Koo, Y.-s.; Shin, W.-c.; Park, H.-j.; Yang, H.-y.; Nam, C.-s. Analysis of Latent Defect Detection Using Sigma Deviation Count Labeling (SDCL). Electronics 2025, 14, 3912. https://doi.org/10.3390/electronics14193912
Koo Y-s, Shin W-c, Park H-j, Yang H-y, Nam C-s. Analysis of Latent Defect Detection Using Sigma Deviation Count Labeling (SDCL). Electronics. 2025; 14(19):3912. https://doi.org/10.3390/electronics14193912
Chicago/Turabian StyleKoo, Yun-su, Woo-chang Shin, Ha-je Park, Hee-yeong Yang, and Choon-sung Nam. 2025. "Analysis of Latent Defect Detection Using Sigma Deviation Count Labeling (SDCL)" Electronics 14, no. 19: 3912. https://doi.org/10.3390/electronics14193912
APA StyleKoo, Y.-s., Shin, W.-c., Park, H.-j., Yang, H.-y., & Nam, C.-s. (2025). Analysis of Latent Defect Detection Using Sigma Deviation Count Labeling (SDCL). Electronics, 14(19), 3912. https://doi.org/10.3390/electronics14193912