A simulation physics-guided neural network for predicting semiconductor structure with few experimental data

Kim, QHwan; Lee, Sung‐Hee; Ma, Ami; Kim, Jaeyoon; Noh, Hyeon‐Kyun; Chang, Kyu Baik; Cheon, Wooyoung; Yi, Shin-Wook; Jeong, Jaehoon; Kim, BongSeok; Kim, Young-Seok; Kim, Dae Sin

doi:10.1016/j.sse.2022.108568

Cited by 4 publications

(2 citation statements)

References 6 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Considering the practical significance of the influencing factors, the trend extrapolation method is used for curve fitting, although the error is small, it is predicted that X1 will have a negative value in 2023-2027, so this model is abandoned and the GM (1,1) model is selected. Use this method to predict each influencing factor index in [2023][2024][2025][2026][2027]. The predicted values are shown in Table 7.…”

Section: Future Tendency Predictionsmentioning

confidence: 99%

An improved machine learning model Shapley value-based to forecast demand for aquatic product supply chain

Huang

2023

Front. Ecol. Evol.

View full text Add to dashboard Cite

Previous machine learning models usually faced the problem of poor performance, especially for aquatic product supply chains. In this study, we proposed a coupling machine learning model Shapely value-based to predict the CCL demand of aquatic products (CCLD-AP). We first select the key impact indicators through the gray correlation degree and finally determine the indicator system. Secondly, gray prediction, principal component regression analysis prediction, and BP neural network models are constructed from the perspective of time series, linear regression and nonlinear, combined with three single forecasts, a combined forecasting model is constructed, the error analysis of all prediction model results shows that the combined prediction results are more accurate. Finally, the trend extrapolation method and time series are combined to predict the independent variable influencing factor value and the CCLD-AP from 2023 to 2027. Our study can provide a reference for the progress of CCLD-AP in ports and their hinterland cities.

show abstract

Section: Future Tendency Predictionsmentioning

confidence: 99%

An improved machine learning model Shapley value-based to forecast demand for aquatic product supply chain

Huang

2023

Front. Ecol. Evol.

View full text Add to dashboard Cite

show abstract

“…The demand of machine learning (ML) in the semiconductor metrology is growing as the complexity of semiconductor structures increases [1][2][3]. As semiconductors become more integrated and perform better with technological advancements, the increasingly fine-tuned nano-structures pose a challenge in metrology [4].…”

Section: Introductionmentioning

confidence: 99%

A few-shot machine learning-based OCD metrology algorithm with anomaly detection and wafer-level data augmentation

Kim,

Chang

et al. 2024

Metrology, Inspection, and Process Control XXXVIII

View full text Add to dashboard Cite

With the increasing complexity of 3-D semiconductor structures, the use of optical critical dimension (OCD) metrology has become a popular solution due to its accuracy and fast inference time. Machine learning has been widely adopted in this field to further improve the efficiency and precision of OCD metrology. Especially for high aspect ratio structures such as DRAM and VNAND, where the required computing power for physical modeling increases exponentially, the importance of machine learning with reference data is crucial. However, one significant challenge of the machine learningbased metrology under rapidly changing process condition is the limitation of available labeled data, which causes overfitting and decreases recipe reliability in the manufacturing process as the cost of wafer consumption increases. To utilize machine learning algorithms in mass production, the development of robust algorithms that can be optimized with few-shot data is required. In this paper, we propose a few-shot machine learning algorithm that includes i) wafer-level statistical information-based data augmentation and ii) anomaly detection to automatically remove data with measurement errors. The proposed algorithm shows superior accuracy, repeatability, and in-wafer uniformity compared to the benchmark algorithm in tests with manufacturing phase data. Additionally, this robustness can be sustained with the minimum amount of data in metrology, as only 9 reference training data are used on three design of experiment (DoE) wafers. The proposed optimized solution is expected to contribute to the reduction of measurement costs and production yields of highly complicated 3D semiconductor structures.

show abstract

A generalizable framework of solution-guided machine learning with application to nanoindentation of free-standing thin films

Wang,

Ying,

Chen

et al. 2024

Thin-Walled Structures

View full text Add to dashboard Cite

A simulation physics-guided neural network for predicting semiconductor structure with few experimental data

Cited by 4 publications

References 6 publications

An improved machine learning model Shapley value-based to forecast demand for aquatic product supply chain

An improved machine learning model Shapley value-based to forecast demand for aquatic product supply chain

A few-shot machine learning-based OCD metrology algorithm with anomaly detection and wafer-level data augmentation

A generalizable framework of solution-guided machine learning with application to nanoindentation of free-standing thin films

Contact Info

Product

Resources

About