Supervised subgraph augmented non-negative matrix factorization for interpretable manufacturing time series data analytics

Hongyue Sun*, Ran Jin, Yuan Luo

*Corresponding author for this work

Research output: Contribution to journalArticle

Abstract

Data analytics has been extensively used for manufacturing time series to reduce process variation and mitigate product defects. However, the majority of data analytics approaches are hard to understand for humans who do not have a data analysis background. Many manufacturing conditions, such as trouble shooting, need situation-dependent responses and are mainly performed by humans. Therefore, it is critical to discover insights from the time series and present those to a human operator in an interpretable format. We propose a novel Supervised Subgraph Augmented Non-negative Matrix Factorization (Super-SANMF) approach to represent and model manufacturing time series. We use a graph representation to approximate a human’s description of time series changing patterns and identify frequent subgraphs as common patterns. The appearances of the subgraphs in the time series are organized in a count matrix, in which each row corresponds to a time series and each column corresponds to a frequent subgraph. Super-SANMF then identifies groups of subgraphs as features that minimize the Kullback–Leibler divergence between measured and approximated matrices. The learned features can yield comparable prediction accuracy (normal or defective) in case studies, compared with the widely used basis expansion approaches (such as spline and wavelet), and are easy for humans to memorize and understand.

Original languageEnglish (US)
Pages (from-to)120-131
Number of pages12
JournalIISE Transactions
Volume52
Issue number1
DOIs
StatePublished - Jan 2 2020

Fingerprint

Factorization
Time series
Splines
Defects

Keywords

  • Interpretable data analytics
  • manufacturing time series
  • subgraph augmented matrix factorization

ASJC Scopus subject areas

  • Industrial and Manufacturing Engineering

Cite this

@article{f8a5ae6fdeae425895056dd8ae8c0438,
title = "Supervised subgraph augmented non-negative matrix factorization for interpretable manufacturing time series data analytics",
abstract = "Data analytics has been extensively used for manufacturing time series to reduce process variation and mitigate product defects. However, the majority of data analytics approaches are hard to understand for humans who do not have a data analysis background. Many manufacturing conditions, such as trouble shooting, need situation-dependent responses and are mainly performed by humans. Therefore, it is critical to discover insights from the time series and present those to a human operator in an interpretable format. We propose a novel Supervised Subgraph Augmented Non-negative Matrix Factorization (Super-SANMF) approach to represent and model manufacturing time series. We use a graph representation to approximate a human’s description of time series changing patterns and identify frequent subgraphs as common patterns. The appearances of the subgraphs in the time series are organized in a count matrix, in which each row corresponds to a time series and each column corresponds to a frequent subgraph. Super-SANMF then identifies groups of subgraphs as features that minimize the Kullback–Leibler divergence between measured and approximated matrices. The learned features can yield comparable prediction accuracy (normal or defective) in case studies, compared with the widely used basis expansion approaches (such as spline and wavelet), and are easy for humans to memorize and understand.",
keywords = "Interpretable data analytics, manufacturing time series, subgraph augmented matrix factorization",
author = "Hongyue Sun and Ran Jin and Yuan Luo",
year = "2020",
month = "1",
day = "2",
doi = "10.1080/24725854.2019.1581389",
language = "English (US)",
volume = "52",
pages = "120--131",
journal = "IISE Transactions",
issn = "2472-5854",
publisher = "Taylor and Francis Ltd.",
number = "1",

}

Supervised subgraph augmented non-negative matrix factorization for interpretable manufacturing time series data analytics. / Sun, Hongyue; Jin, Ran; Luo, Yuan.

In: IISE Transactions, Vol. 52, No. 1, 02.01.2020, p. 120-131.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Supervised subgraph augmented non-negative matrix factorization for interpretable manufacturing time series data analytics

AU - Sun, Hongyue

AU - Jin, Ran

AU - Luo, Yuan

PY - 2020/1/2

Y1 - 2020/1/2

N2 - Data analytics has been extensively used for manufacturing time series to reduce process variation and mitigate product defects. However, the majority of data analytics approaches are hard to understand for humans who do not have a data analysis background. Many manufacturing conditions, such as trouble shooting, need situation-dependent responses and are mainly performed by humans. Therefore, it is critical to discover insights from the time series and present those to a human operator in an interpretable format. We propose a novel Supervised Subgraph Augmented Non-negative Matrix Factorization (Super-SANMF) approach to represent and model manufacturing time series. We use a graph representation to approximate a human’s description of time series changing patterns and identify frequent subgraphs as common patterns. The appearances of the subgraphs in the time series are organized in a count matrix, in which each row corresponds to a time series and each column corresponds to a frequent subgraph. Super-SANMF then identifies groups of subgraphs as features that minimize the Kullback–Leibler divergence between measured and approximated matrices. The learned features can yield comparable prediction accuracy (normal or defective) in case studies, compared with the widely used basis expansion approaches (such as spline and wavelet), and are easy for humans to memorize and understand.

AB - Data analytics has been extensively used for manufacturing time series to reduce process variation and mitigate product defects. However, the majority of data analytics approaches are hard to understand for humans who do not have a data analysis background. Many manufacturing conditions, such as trouble shooting, need situation-dependent responses and are mainly performed by humans. Therefore, it is critical to discover insights from the time series and present those to a human operator in an interpretable format. We propose a novel Supervised Subgraph Augmented Non-negative Matrix Factorization (Super-SANMF) approach to represent and model manufacturing time series. We use a graph representation to approximate a human’s description of time series changing patterns and identify frequent subgraphs as common patterns. The appearances of the subgraphs in the time series are organized in a count matrix, in which each row corresponds to a time series and each column corresponds to a frequent subgraph. Super-SANMF then identifies groups of subgraphs as features that minimize the Kullback–Leibler divergence between measured and approximated matrices. The learned features can yield comparable prediction accuracy (normal or defective) in case studies, compared with the widely used basis expansion approaches (such as spline and wavelet), and are easy for humans to memorize and understand.

KW - Interpretable data analytics

KW - manufacturing time series

KW - subgraph augmented matrix factorization

UR - http://www.scopus.com/inward/record.url?scp=85065327819&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85065327819&partnerID=8YFLogxK

U2 - 10.1080/24725854.2019.1581389

DO - 10.1080/24725854.2019.1581389

M3 - Article

AN - SCOPUS:85065327819

VL - 52

SP - 120

EP - 131

JO - IISE Transactions

JF - IISE Transactions

SN - 2472-5854

IS - 1

ER -