Silicon Wafer Map Defect Classification Using Deep Convolutional Neural Network With Data Augmentation

Shawon, Ashadullah; Faruk, Omar; Habib, Masrur Bin; Khan, Abdullah

doi:10.1109/iccc47050.2019.9064029

Cited by 16 publications

(8 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To show the effectiveness of the proposed model in WMDPI, we compare the proposed model with previously published results with different input image resolutions. For 26 × 26 input image resolution, the proposed model is compared with Shawon et al 16 The train and test samples used in Shawon et al 16 are 12,730 and 705, respectively. The proposed model uses the same training and test samples and repeats 10 times to obtain performance values.…”

Section: The Opt-resdcnn Model Comparison Resultsmentioning

confidence: 99%

“…Shawon et al 16 proposed a DCNN architecture with a data augmentation technique to solve the data balance problem and achieve high accuracy. Yu et al 17 presented a hybrid DL model called stacked convolutional sparse denoising autoencoder (SCSDAE).…”

Section: Related Workmentioning

confidence: 99%

“…Many models have been proposed to deal with data imbalance problems. 16,17,[31][32][33][34] The CAE model is used for data generation purposes to solve the data imbalance issue in this study. The CAE model is an unsupervised DL model.…”

Section: Data Balancing With Caementioning

confidence: 99%

“…The proposed model called Opt-ResDCNN is a unique model which combines the concept from the DCNN model and residual blocks. Previously, Shawon et al 16 proposed a DCNN model with wafer map pattern recognition accuracy of 99.26% for the WM-811K dataset. This study tries to improve WMDPI and classification based on the DCNN idea and add residual blocks to connect more convolutional layers to increase the neural network's depth.…”

Section: Proposed DL Classifier: Opt-resdcnn Modelmentioning

confidence: 99%

See 3 more Smart Citations

A Deep convolutional neural network with residual blocks for wafer map defect pattern recognition

Wang

Chou

Amogne

2021

Quality & Reliability Eng

View full text Add to dashboard Cite

Different deep convolution neural network (DCNN) models have been proposed for wafer map pattern identification and classification tasks in previous studies. However, factors such as input image resolution effect on the classification performance of the proposed models and class imbalance in the training set after splitting the data into training and test sets have not been considered in the previous studies. We propose a DCNN model with residual blocks, called the Opt-ResDCNN model, for wafer map defect pattern classification by considering 26 × 26, 64 × 64, 96 × 96, and 256 × 256 input images and class imbalance issues. The model with a balance function can improve the performance.We compare the proposed model with the published defect pattern classification models in terms of accuracy, precision, recall, and F1 value. Using a publicly available wafer map data set (WM-811K), the proposed method on the four different resolutions can obtain an excellent average accuracy, precision, recall, and F1 score. Regarding accuracy, the proposed model results are 99.90%, 99.86%, 90.28%, 98.88%, respectively. These results are better than the published papers.

show abstract

Section: The Opt-resdcnn Model Comparison Resultsmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Section: Data Balancing With Caementioning

confidence: 99%

Section: Proposed DL Classifier: Opt-resdcnn Modelmentioning

confidence: 99%

See 2 more Smart Citations

A Deep convolutional neural network with residual blocks for wafer map defect pattern recognition

Wang

Chou

Amogne

2021

Quality & Reliability Eng

View full text Add to dashboard Cite

show abstract

“…Then, CNN and extreme gradient boosting methods are employed for wafer map retrieval and defect pattern classification. Shawon et al [36] also modified the CNN architecture to improve the classification performance and used data augmentation techniques to solve the data imbalance problem. Nakazawa and Kulkarni [37] proposed a deep convolutional encoder-decoder neural network architecture for detecting wafer map defect patterns, as well as segmentation.…”

Section: Related Workmentioning

confidence: 99%

Unsupervised Pre-Training of Imbalanced Data for Identification of Wafer Map Defect Patterns

et al. 2021

View full text Add to dashboard Cite

Visual defect inspection and classification are significant steps of most manufacturing processes in the semiconductor and electronics industries. Known and unknown defects on wafer maps tend to cluster, and these spatial patterns provide valuable process information for supporting manufacturing in determining the root causes of abnormal processes. In previous studies, data augmentation-based deep learning (DL) techniques were most commonly used for the identification of wafer map defect patterns (WMDP). Data augmentation is an effective technique for improving the accuracy of modern image classifiers. However, current data augmentation implementations were manually designed for the WMDP problem. In this study, we propose a DL-based method with automatic data augmentation for the WMDP task. Basically, it focuses on learning effective discriminative features, from wafer maps, through a deep network structure. The network consists of a convolution-based variational autoencoder (CVAE) sequentially. First, we pre-trained the CVAE on large training data in an unsupervised manner. Second, we fine-tuned the encoder of the CVAE, which was followed by a neural network (NN) classifier, in a supervised manner. Additionally, we describe a simple procedure for automatically searching for improved data augmentation policies. The policy mainly consists of five image processing functions: rotation, flipping, shifting, shearing range, and zooming. The effectiveness of the proposed method was demonstrated through experimental results obtained from a simulation dataset and a real-world wafer map dataset (WM-811K). This study provides guidance for the application of deep learning in semiconductor manufacturing processes to improve product quality and yield.INDEX TERMS Classification, convolutional variational autoencoder, deep learning, imbalanced data, neural network, unsupervised pre-training, variational autoencoder, wafer map defect patterns.

show abstract

Deep Convolutional Neural Networks with Residual Blocks for Wafer Map Defect Pattern Recognition

Amogne

Wang

Chou

2021

Advances in Computational Intelligence

View full text Add to dashboard Cite

Silicon Wafer Map Defect Classification Using Deep Convolutional Neural Network With Data Augmentation

Cited by 16 publications

References 9 publications

A Deep convolutional neural network with residual blocks for wafer map defect pattern recognition

A Deep convolutional neural network with residual blocks for wafer map defect pattern recognition

Unsupervised Pre-Training of Imbalanced Data for Identification of Wafer Map Defect Patterns

Deep Convolutional Neural Networks with Residual Blocks for Wafer Map Defect Pattern Recognition

Contact Info

Product

Resources

About