Online Bivariate Outlier Detection in Final Test Using Kernel Density Estimation

Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

67 Downloads (Pure)

Abstract

In parametric IC testing, outlier detection is applied to filter out potential unreliable devices. Most outlier detection methods are used in an offline setting and hence are not applicable to Final Test, where immediate pass/fail decisions are required. Therefore, we developed a new bivariate online outlier detection method that is applicable to Final Test without making assumptions about a specific form of relations between two test parameters. An acceptance region is constructed using kernel density estimation. We use a grid discretization in order to enable a fast outlier decision. After each accepted device the grid is updated, hence the method is able to adapt to shifting measurements.
Original languageUndefined
Title of host publicationIEEE Defect and Adaptive Test Analysis Workshop
Place of PublicationUSA
PublisherIEEE Computer Society
Pages-
Number of pages6
ISBN (Print)not assigned
Publication statusPublished - 22 Sep 2011

Publication series

Name
PublisherIEEE Computer Society

Keywords

  • EWI-22580
  • Kernel Density
  • METIS-289798
  • IR-83435
  • Adaptive
  • Outlier Detection

Cite this

Bossers, H. C. M., Hurink, J. L., & Smit, G. J. M. (2011). Online Bivariate Outlier Detection in Final Test Using Kernel Density Estimation. In IEEE Defect and Adaptive Test Analysis Workshop (pp. -). USA: IEEE Computer Society.