Self-Supervised Representation Learning for Wafer Bin Map Defect Pattern Classification

Hyungu Kahng; Seoung Bum Kim

doi:10.1109/TSM.2020.3038165

Self-Supervised Representation Learning for Wafer Bin Map Defect Pattern Classification

Hyungu Kahng
, Seoung Bum Kim^*

^*Corresponding author for this work

Research output: Contribution to journal › Article › peer-review

65 Citations (Scopus)

Abstract

Automatic identification of defect patterns in wafer bin maps (WBMs) stands as a challenging problem for the semiconductor manufacturing industry. Deep convolutional neural networks have recently shown decent progress in learning spatial patterns in WBMs, but only at the expense of explicit manual supervision. Unfortunately, a clean set of labeled WBM samples is often limited in both size and quality, especially during rapid process development or early production stages. In this study, we propose a self-supervised learning framework that makes the most out of unlabeled data to learn beforehand rich visual representations for data-efficient WBM defect pattern classification. After self-supervised pre-training based on noise-contrastive estimation, the network is fine-tuned on the available labeled data to classify WBM defect patterns. We argue that self-supervised pre-training with a vast amount of unlabeled data substantially improves classification performance when labels are scarce. We demonstrate the effectiveness of our work on a real-world public WBM dataset, WM-811K. The code is available at https://github.com/hgkahng/WaPIRL.

Original language	English
Article number	9260238
Pages (from-to)	74-86
Number of pages	13
Journal	IEEE Transactions on Semiconductor Manufacturing
Volume	34
Issue number	1
DOIs	https://doi.org/10.1109/TSM.2020.3038165
Publication status	Published - 2021 Feb

Bibliographical note

Funding Information:
Manuscript received September 9, 2020; revised October 9, 2020 and October 31, 2020; accepted November 11, 2020. Date of publication November 16, 2020; date of current version February 3, 2021. This work was supported in part by the Korea Institute for Advancement of Technology (KIAT) Grant funded by the Korean Government (MOTIE) under Grant P0008691 (Competency Development Program for Industry Specialists), and in part by the National Research Foundation of Korea Grant funded by the Korean Government (MSIT) under Grant NRF-2019R1A4A1024732. (Corresponding author: Seoung Bum Kim.) The authors are with the School of Industrial Management Engineering, Korea University, Seoul 02841, South Korea (e-mail: [email protected]; [email protected]).

Publisher Copyright:
© 1988-2012 IEEE.

Keywords

Wafer bin map defect pattern classification
contrastive learning
convolutional neural networks
self-supervised learning
semiconductor manufacturing
unsupervised learning

ASJC Scopus subject areas

Electronic, Optical and Magnetic Materials
Condensed Matter Physics
Industrial and Manufacturing Engineering
Electrical and Electronic Engineering

Access to Document

10.1109/TSM.2020.3038165

Cite this

@article{1d83149f7e0f43b6b707c3eac0d806a8,

title = "Self-Supervised Representation Learning for Wafer Bin Map Defect Pattern Classification",

abstract = "Automatic identification of defect patterns in wafer bin maps (WBMs) stands as a challenging problem for the semiconductor manufacturing industry. Deep convolutional neural networks have recently shown decent progress in learning spatial patterns in WBMs, but only at the expense of explicit manual supervision. Unfortunately, a clean set of labeled WBM samples is often limited in both size and quality, especially during rapid process development or early production stages. In this study, we propose a self-supervised learning framework that makes the most out of unlabeled data to learn beforehand rich visual representations for data-efficient WBM defect pattern classification. After self-supervised pre-training based on noise-contrastive estimation, the network is fine-tuned on the available labeled data to classify WBM defect patterns. We argue that self-supervised pre-training with a vast amount of unlabeled data substantially improves classification performance when labels are scarce. We demonstrate the effectiveness of our work on a real-world public WBM dataset, WM-811K. The code is available at https://github.com/hgkahng/WaPIRL.",

keywords = "Wafer bin map defect pattern classification, contrastive learning, convolutional neural networks, self-supervised learning, semiconductor manufacturing, unsupervised learning",

author = "Hyungu Kahng and Kim, \{Seoung Bum\}",

note = "Funding Information: Manuscript received September 9, 2020; revised October 9, 2020 and October 31, 2020; accepted November 11, 2020. Date of publication November 16, 2020; date of current version February 3, 2021. This work was supported in part by the Korea Institute for Advancement of Technology (KIAT) Grant funded by the Korean Government (MOTIE) under Grant P0008691 (Competency Development Program for Industry Specialists), and in part by the National Research Foundation of Korea Grant funded by the Korean Government (MSIT) under Grant NRF-2019R1A4A1024732. (Corresponding author: Seoung Bum Kim.) The authors are with the School of Industrial Management Engineering, Korea University, Seoul 02841, South Korea (e-mail: [email protected]; [email protected]). Publisher Copyright: {\textcopyright} 1988-2012 IEEE.",

year = "2021",

month = feb,

doi = "10.1109/TSM.2020.3038165",

language = "English",

volume = "34",

pages = "74--86",

journal = "IEEE Transactions on Semiconductor Manufacturing",

issn = "0894-6507",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

number = "1",

}

TY - JOUR

T1 - Self-Supervised Representation Learning for Wafer Bin Map Defect Pattern Classification

AU - Kahng, Hyungu

AU - Kim, Seoung Bum

N1 - Funding Information: Manuscript received September 9, 2020; revised October 9, 2020 and October 31, 2020; accepted November 11, 2020. Date of publication November 16, 2020; date of current version February 3, 2021. This work was supported in part by the Korea Institute for Advancement of Technology (KIAT) Grant funded by the Korean Government (MOTIE) under Grant P0008691 (Competency Development Program for Industry Specialists), and in part by the National Research Foundation of Korea Grant funded by the Korean Government (MSIT) under Grant NRF-2019R1A4A1024732. (Corresponding author: Seoung Bum Kim.) The authors are with the School of Industrial Management Engineering, Korea University, Seoul 02841, South Korea (e-mail: [email protected]; [email protected]). Publisher Copyright: © 1988-2012 IEEE.

PY - 2021/2

Y1 - 2021/2

N2 - Automatic identification of defect patterns in wafer bin maps (WBMs) stands as a challenging problem for the semiconductor manufacturing industry. Deep convolutional neural networks have recently shown decent progress in learning spatial patterns in WBMs, but only at the expense of explicit manual supervision. Unfortunately, a clean set of labeled WBM samples is often limited in both size and quality, especially during rapid process development or early production stages. In this study, we propose a self-supervised learning framework that makes the most out of unlabeled data to learn beforehand rich visual representations for data-efficient WBM defect pattern classification. After self-supervised pre-training based on noise-contrastive estimation, the network is fine-tuned on the available labeled data to classify WBM defect patterns. We argue that self-supervised pre-training with a vast amount of unlabeled data substantially improves classification performance when labels are scarce. We demonstrate the effectiveness of our work on a real-world public WBM dataset, WM-811K. The code is available at https://github.com/hgkahng/WaPIRL.

AB - Automatic identification of defect patterns in wafer bin maps (WBMs) stands as a challenging problem for the semiconductor manufacturing industry. Deep convolutional neural networks have recently shown decent progress in learning spatial patterns in WBMs, but only at the expense of explicit manual supervision. Unfortunately, a clean set of labeled WBM samples is often limited in both size and quality, especially during rapid process development or early production stages. In this study, we propose a self-supervised learning framework that makes the most out of unlabeled data to learn beforehand rich visual representations for data-efficient WBM defect pattern classification. After self-supervised pre-training based on noise-contrastive estimation, the network is fine-tuned on the available labeled data to classify WBM defect patterns. We argue that self-supervised pre-training with a vast amount of unlabeled data substantially improves classification performance when labels are scarce. We demonstrate the effectiveness of our work on a real-world public WBM dataset, WM-811K. The code is available at https://github.com/hgkahng/WaPIRL.

KW - Wafer bin map defect pattern classification

KW - contrastive learning

KW - convolutional neural networks

KW - self-supervised learning

KW - semiconductor manufacturing

KW - unsupervised learning

UR - https://www.scopus.com/pages/publications/85098753376

U2 - 10.1109/TSM.2020.3038165

DO - 10.1109/TSM.2020.3038165

M3 - Article

AN - SCOPUS:85098753376

SN - 0894-6507

VL - 34

SP - 74

EP - 86

JO - IEEE Transactions on Semiconductor Manufacturing

JF - IEEE Transactions on Semiconductor Manufacturing

IS - 1

M1 - 9260238

ER -

Self-Supervised Representation Learning for Wafer Bin Map Defect Pattern Classification

Abstract

Bibliographical note

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this