Reducing correspondence ambiguity in loosely labeled training data

Jacobus J Barnard, Quanfu Fan

Research output: Chapter in Book/Report/Conference proceedingConference contribution

6 Citations (Scopus)

Abstract

We develop an approach to reduce correspondence ambiguity in training data where data items are associated with sets of plausible labels. Our domain is images annotated with keywords where it is not known which part of the image a keyword refers to. In contrast to earlier approaches that build predictive models or classifiers despite the ambiguity, we argue that that it is better to first address the correspondence ambiguity, and then build more complex models from the improved training data. This addresses difficulties of fitting complex models in the face of ambiguity while exploiting all the constraints available from the training data. We contribute a simple and flexible formulation of the problem, and show results validated by a recently developed comprehensive evaluation data set and corresponding evaluation methodology.

Original languageEnglish (US)
Title of host publicationProceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition
DOIs
StatePublished - 2007
Event2007 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR'07 - Minneapolis, MN, United States
Duration: Jun 17 2007Jun 22 2007

Other

Other2007 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR'07
CountryUnited States
CityMinneapolis, MN
Period6/17/076/22/07

Fingerprint

Labels
Classifiers

ASJC Scopus subject areas

  • Electrical and Electronic Engineering
  • Computer Vision and Pattern Recognition
  • Software
  • Control and Systems Engineering

Cite this

Barnard, J. J., & Fan, Q. (2007). Reducing correspondence ambiguity in loosely labeled training data. In Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition [4270249] https://doi.org/10.1109/CVPR.2007.383224

Reducing correspondence ambiguity in loosely labeled training data. / Barnard, Jacobus J; Fan, Quanfu.

Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition. 2007. 4270249.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Barnard, JJ & Fan, Q 2007, Reducing correspondence ambiguity in loosely labeled training data. in Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition., 4270249, 2007 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR'07, Minneapolis, MN, United States, 6/17/07. https://doi.org/10.1109/CVPR.2007.383224
Barnard JJ, Fan Q. Reducing correspondence ambiguity in loosely labeled training data. In Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition. 2007. 4270249 https://doi.org/10.1109/CVPR.2007.383224
Barnard, Jacobus J ; Fan, Quanfu. / Reducing correspondence ambiguity in loosely labeled training data. Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition. 2007.
@inproceedings{fe66dbeb79fc470781b9012a6b608a44,
title = "Reducing correspondence ambiguity in loosely labeled training data",
abstract = "We develop an approach to reduce correspondence ambiguity in training data where data items are associated with sets of plausible labels. Our domain is images annotated with keywords where it is not known which part of the image a keyword refers to. In contrast to earlier approaches that build predictive models or classifiers despite the ambiguity, we argue that that it is better to first address the correspondence ambiguity, and then build more complex models from the improved training data. This addresses difficulties of fitting complex models in the face of ambiguity while exploiting all the constraints available from the training data. We contribute a simple and flexible formulation of the problem, and show results validated by a recently developed comprehensive evaluation data set and corresponding evaluation methodology.",
author = "Barnard, {Jacobus J} and Quanfu Fan",
year = "2007",
doi = "10.1109/CVPR.2007.383224",
language = "English (US)",
isbn = "1424411807",
booktitle = "Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition",

}

TY - GEN

T1 - Reducing correspondence ambiguity in loosely labeled training data

AU - Barnard, Jacobus J

AU - Fan, Quanfu

PY - 2007

Y1 - 2007

N2 - We develop an approach to reduce correspondence ambiguity in training data where data items are associated with sets of plausible labels. Our domain is images annotated with keywords where it is not known which part of the image a keyword refers to. In contrast to earlier approaches that build predictive models or classifiers despite the ambiguity, we argue that that it is better to first address the correspondence ambiguity, and then build more complex models from the improved training data. This addresses difficulties of fitting complex models in the face of ambiguity while exploiting all the constraints available from the training data. We contribute a simple and flexible formulation of the problem, and show results validated by a recently developed comprehensive evaluation data set and corresponding evaluation methodology.

AB - We develop an approach to reduce correspondence ambiguity in training data where data items are associated with sets of plausible labels. Our domain is images annotated with keywords where it is not known which part of the image a keyword refers to. In contrast to earlier approaches that build predictive models or classifiers despite the ambiguity, we argue that that it is better to first address the correspondence ambiguity, and then build more complex models from the improved training data. This addresses difficulties of fitting complex models in the face of ambiguity while exploiting all the constraints available from the training data. We contribute a simple and flexible formulation of the problem, and show results validated by a recently developed comprehensive evaluation data set and corresponding evaluation methodology.

UR - http://www.scopus.com/inward/record.url?scp=34948849778&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=34948849778&partnerID=8YFLogxK

U2 - 10.1109/CVPR.2007.383224

DO - 10.1109/CVPR.2007.383224

M3 - Conference contribution

AN - SCOPUS:34948849778

SN - 1424411807

SN - 9781424411801

BT - Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition

ER -