A general framework for predictive tensor modeling with domain knowledge

Yada Zhu, Jingrui He, Richard D. Lawrence

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

In many real applications such as virtual metrology in semiconductor manufacturing, face recognition, and gait recognition in computer vision, the input data is naturally expressed as tensors or multi-dimensional arrays. Furthermore, in addition to the known label information, domain knowledge can often be obtained from various sources, e.g., multiple domain experts. To address such problems, in this paper, we propose a general optimization framework for dealing with tensor inputs while taking into consideration domain knowledge. To be specific, our framework is based on a linear model, and we obtain the weight tensor in a hierarchical way—first approximate it by a low-rank tensor, and then estimate the low-rank approximation using the domain knowledge from various sources. This is motivated by wafer quality prediction in semiconductor manufacturing. We also propose an effective algorithm named H-MOTE for solving this framework, which is guaranteed to converge. For each iteration, the time complexity of H-MOTE is linear with respect to the number of examples as well as the size of the weight tensor. Therefore, H-MOTE is scalable to large-scale problems. Experimental results show that H-MOTE outperforms state-of-the-art techniques on both synthetic and real data sets.

Original languageEnglish (US)
Pages (from-to)1709-1732
Number of pages24
JournalData Mining and Knowledge Discovery
Volume29
Issue number6
DOIs
StatePublished - Nov 12 2014

Fingerprint

Tensors
Semiconductor materials
Face recognition
Computer vision
Labels

Keywords

  • Algorithms
  • Classifier design and evaluation
  • Data mining
  • Semiconductor manufacturing
  • Tensor
  • Virtual metrology
  • Wafer quality

ASJC Scopus subject areas

  • Information Systems
  • Computer Science Applications
  • Computer Networks and Communications

Cite this

A general framework for predictive tensor modeling with domain knowledge. / Zhu, Yada; He, Jingrui; Lawrence, Richard D.

In: Data Mining and Knowledge Discovery, Vol. 29, No. 6, 12.11.2014, p. 1709-1732.

Research output: Contribution to journalArticle

@article{0194ad37f42248f6b52f146f63eba9f0,
title = "A general framework for predictive tensor modeling with domain knowledge",
abstract = "In many real applications such as virtual metrology in semiconductor manufacturing, face recognition, and gait recognition in computer vision, the input data is naturally expressed as tensors or multi-dimensional arrays. Furthermore, in addition to the known label information, domain knowledge can often be obtained from various sources, e.g., multiple domain experts. To address such problems, in this paper, we propose a general optimization framework for dealing with tensor inputs while taking into consideration domain knowledge. To be specific, our framework is based on a linear model, and we obtain the weight tensor in a hierarchical way—first approximate it by a low-rank tensor, and then estimate the low-rank approximation using the domain knowledge from various sources. This is motivated by wafer quality prediction in semiconductor manufacturing. We also propose an effective algorithm named H-MOTE for solving this framework, which is guaranteed to converge. For each iteration, the time complexity of H-MOTE is linear with respect to the number of examples as well as the size of the weight tensor. Therefore, H-MOTE is scalable to large-scale problems. Experimental results show that H-MOTE outperforms state-of-the-art techniques on both synthetic and real data sets.",
keywords = "Algorithms, Classifier design and evaluation, Data mining, Semiconductor manufacturing, Tensor, Virtual metrology, Wafer quality",
author = "Yada Zhu and Jingrui He and Lawrence, {Richard D.}",
year = "2014",
month = "11",
day = "12",
doi = "10.1007/s10618-014-0392-8",
language = "English (US)",
volume = "29",
pages = "1709--1732",
journal = "Data Mining and Knowledge Discovery",
issn = "1384-5810",
publisher = "Springer Netherlands",
number = "6",

}

TY - JOUR

T1 - A general framework for predictive tensor modeling with domain knowledge

AU - Zhu, Yada

AU - He, Jingrui

AU - Lawrence, Richard D.

PY - 2014/11/12

Y1 - 2014/11/12

N2 - In many real applications such as virtual metrology in semiconductor manufacturing, face recognition, and gait recognition in computer vision, the input data is naturally expressed as tensors or multi-dimensional arrays. Furthermore, in addition to the known label information, domain knowledge can often be obtained from various sources, e.g., multiple domain experts. To address such problems, in this paper, we propose a general optimization framework for dealing with tensor inputs while taking into consideration domain knowledge. To be specific, our framework is based on a linear model, and we obtain the weight tensor in a hierarchical way—first approximate it by a low-rank tensor, and then estimate the low-rank approximation using the domain knowledge from various sources. This is motivated by wafer quality prediction in semiconductor manufacturing. We also propose an effective algorithm named H-MOTE for solving this framework, which is guaranteed to converge. For each iteration, the time complexity of H-MOTE is linear with respect to the number of examples as well as the size of the weight tensor. Therefore, H-MOTE is scalable to large-scale problems. Experimental results show that H-MOTE outperforms state-of-the-art techniques on both synthetic and real data sets.

AB - In many real applications such as virtual metrology in semiconductor manufacturing, face recognition, and gait recognition in computer vision, the input data is naturally expressed as tensors or multi-dimensional arrays. Furthermore, in addition to the known label information, domain knowledge can often be obtained from various sources, e.g., multiple domain experts. To address such problems, in this paper, we propose a general optimization framework for dealing with tensor inputs while taking into consideration domain knowledge. To be specific, our framework is based on a linear model, and we obtain the weight tensor in a hierarchical way—first approximate it by a low-rank tensor, and then estimate the low-rank approximation using the domain knowledge from various sources. This is motivated by wafer quality prediction in semiconductor manufacturing. We also propose an effective algorithm named H-MOTE for solving this framework, which is guaranteed to converge. For each iteration, the time complexity of H-MOTE is linear with respect to the number of examples as well as the size of the weight tensor. Therefore, H-MOTE is scalable to large-scale problems. Experimental results show that H-MOTE outperforms state-of-the-art techniques on both synthetic and real data sets.

KW - Algorithms

KW - Classifier design and evaluation

KW - Data mining

KW - Semiconductor manufacturing

KW - Tensor

KW - Virtual metrology

KW - Wafer quality

UR - http://www.scopus.com/inward/record.url?scp=84942503850&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84942503850&partnerID=8YFLogxK

U2 - 10.1007/s10618-014-0392-8

DO - 10.1007/s10618-014-0392-8

M3 - Article

VL - 29

SP - 1709

EP - 1732

JO - Data Mining and Knowledge Discovery

JF - Data Mining and Knowledge Discovery

SN - 1384-5810

IS - 6

ER -