Probabilistic local reconstruction for k-NN regression and its application to virtual metrology in semiconductor manufacturing

Seung kyung Lee, Pilsung Kang, Sungzoon Cho

Research output: Contribution to journalArticle

18 Citations (Scopus)

Abstract

The ". locally linear reconstruction" (LLR) provides a principled and k-insensitive way to determine the weights of k-nearest neighbor (k-NN) learning. LLR, however, does not provide a confidence interval for the k neighbors-based reconstruction of a query point, which is required in many real application domains. Moreover, its fixed linear structure makes the local reconstruction model unstable, resulting in performance fluctuation for regressions under different k values. Therefore, we propose a probabilistic local reconstruction (PLR) as an extended version of LLR in the k-NN regression. First, we probabilistically capture the reconstruction uncertainty by incorporating Gaussian regularization prior into the reconstruction model. This prevents over-fitting when there are no informative neighbors in the local reconstruction. We then project data into a higher dimensional feature space to capture the non-linear relationship between neighbors and a query point when a value of k is large. Preliminary experimental results demonstrated that the proposed Bayesian kernel treatment improves accuracy and k-invariance. Moreover, from the experiment on a real virtual metrology data set in the semiconductor manufacturing, it was found that the uncertainty information on the prediction outcomes provided by PLR supports more appropriate decision making.

Original languageEnglish
Pages (from-to)427-439
Number of pages13
JournalNeurocomputing
Volume131
DOIs
Publication statusPublished - 2014 May 5
Externally publishedYes

Fingerprint

Semiconductors
Uncertainty
Semiconductor materials
Invariance
Decision Making
Decision making
Learning
Confidence Intervals
Weights and Measures
Experiments
Datasets

Keywords

  • Bayesian kernel model
  • K-NN regression
  • Locally linear reconstruction

ASJC Scopus subject areas

  • Computer Science Applications
  • Cognitive Neuroscience
  • Artificial Intelligence

Cite this

Probabilistic local reconstruction for k-NN regression and its application to virtual metrology in semiconductor manufacturing. / Lee, Seung kyung; Kang, Pilsung; Cho, Sungzoon.

In: Neurocomputing, Vol. 131, 05.05.2014, p. 427-439.

Research output: Contribution to journalArticle

@article{5eebacc530e04cfaaffc2b032b9d0b58,
title = "Probabilistic local reconstruction for k-NN regression and its application to virtual metrology in semiconductor manufacturing",
abstract = "The {"}. locally linear reconstruction{"} (LLR) provides a principled and k-insensitive way to determine the weights of k-nearest neighbor (k-NN) learning. LLR, however, does not provide a confidence interval for the k neighbors-based reconstruction of a query point, which is required in many real application domains. Moreover, its fixed linear structure makes the local reconstruction model unstable, resulting in performance fluctuation for regressions under different k values. Therefore, we propose a probabilistic local reconstruction (PLR) as an extended version of LLR in the k-NN regression. First, we probabilistically capture the reconstruction uncertainty by incorporating Gaussian regularization prior into the reconstruction model. This prevents over-fitting when there are no informative neighbors in the local reconstruction. We then project data into a higher dimensional feature space to capture the non-linear relationship between neighbors and a query point when a value of k is large. Preliminary experimental results demonstrated that the proposed Bayesian kernel treatment improves accuracy and k-invariance. Moreover, from the experiment on a real virtual metrology data set in the semiconductor manufacturing, it was found that the uncertainty information on the prediction outcomes provided by PLR supports more appropriate decision making.",
keywords = "Bayesian kernel model, K-NN regression, Locally linear reconstruction",
author = "Lee, {Seung kyung} and Pilsung Kang and Sungzoon Cho",
year = "2014",
month = "5",
day = "5",
doi = "10.1016/j.neucom.2013.10.001",
language = "English",
volume = "131",
pages = "427--439",
journal = "Neurocomputing",
issn = "0925-2312",
publisher = "Elsevier",

}

TY - JOUR

T1 - Probabilistic local reconstruction for k-NN regression and its application to virtual metrology in semiconductor manufacturing

AU - Lee, Seung kyung

AU - Kang, Pilsung

AU - Cho, Sungzoon

PY - 2014/5/5

Y1 - 2014/5/5

N2 - The ". locally linear reconstruction" (LLR) provides a principled and k-insensitive way to determine the weights of k-nearest neighbor (k-NN) learning. LLR, however, does not provide a confidence interval for the k neighbors-based reconstruction of a query point, which is required in many real application domains. Moreover, its fixed linear structure makes the local reconstruction model unstable, resulting in performance fluctuation for regressions under different k values. Therefore, we propose a probabilistic local reconstruction (PLR) as an extended version of LLR in the k-NN regression. First, we probabilistically capture the reconstruction uncertainty by incorporating Gaussian regularization prior into the reconstruction model. This prevents over-fitting when there are no informative neighbors in the local reconstruction. We then project data into a higher dimensional feature space to capture the non-linear relationship between neighbors and a query point when a value of k is large. Preliminary experimental results demonstrated that the proposed Bayesian kernel treatment improves accuracy and k-invariance. Moreover, from the experiment on a real virtual metrology data set in the semiconductor manufacturing, it was found that the uncertainty information on the prediction outcomes provided by PLR supports more appropriate decision making.

AB - The ". locally linear reconstruction" (LLR) provides a principled and k-insensitive way to determine the weights of k-nearest neighbor (k-NN) learning. LLR, however, does not provide a confidence interval for the k neighbors-based reconstruction of a query point, which is required in many real application domains. Moreover, its fixed linear structure makes the local reconstruction model unstable, resulting in performance fluctuation for regressions under different k values. Therefore, we propose a probabilistic local reconstruction (PLR) as an extended version of LLR in the k-NN regression. First, we probabilistically capture the reconstruction uncertainty by incorporating Gaussian regularization prior into the reconstruction model. This prevents over-fitting when there are no informative neighbors in the local reconstruction. We then project data into a higher dimensional feature space to capture the non-linear relationship between neighbors and a query point when a value of k is large. Preliminary experimental results demonstrated that the proposed Bayesian kernel treatment improves accuracy and k-invariance. Moreover, from the experiment on a real virtual metrology data set in the semiconductor manufacturing, it was found that the uncertainty information on the prediction outcomes provided by PLR supports more appropriate decision making.

KW - Bayesian kernel model

KW - K-NN regression

KW - Locally linear reconstruction

UR - http://www.scopus.com/inward/record.url?scp=84894105629&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84894105629&partnerID=8YFLogxK

U2 - 10.1016/j.neucom.2013.10.001

DO - 10.1016/j.neucom.2013.10.001

M3 - Article

AN - SCOPUS:84894105629

VL - 131

SP - 427

EP - 439

JO - Neurocomputing

JF - Neurocomputing

SN - 0925-2312

ER -