A tale of two classifiers: SNoW vs. SVM in visual recognition

Ming Hsuan Yang, Dan Roth, Narendra Ahuja

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

8 Citations (Scopus)

Abstract

Numerous statistical learning methods have been developed for visual recognition tasks. Few attempts, however, have been made to address theoretical issues, and in particular, study the suitability of different learning algorithms for visual recognition. Large margin classifiers, such as SNoW and SVM, have recently demonstrated their success in object detection and recognition. In this paper, we present a theoretical account of these two learning approaches, and their suitability to visual recognition. Using tools from computational learning theory, we show that the main difference between the generalization bounds of SVM and SNoW depends on the properties of the data. We argue that learning problems in the visual domain have sparseness characteristics and exhibit them by analyzing data taken from face detection experiments. Experimental results exhibit good generalization and robustness properties of the SNoW-based method, and conform to the theoretical analysis.

Original languageEnglish
Title of host publicationComputer Vision - ECCV 2002 - 7th European Conference on Computer Vision, Proceedings
EditorsMads Nielsen, Anders Heyden, Gunnar Sparr, Peter Johansen
PublisherSpringer Verlag
Pages685-699
Number of pages15
ISBN (Electronic)9783540437482
Publication statusPublished - 2002 Jan 1
Event7th European Conference on Computer Vision, ECCV 2002 - Copenhagen, Denmark
Duration: 2002 May 282002 May 31

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume2353
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference7th European Conference on Computer Vision, ECCV 2002
CountryDenmark
CityCopenhagen
Period02/5/2802/5/31

Fingerprint

Object recognition
Face recognition
Learning algorithms
Classifiers
Classifier
Experiments
Computational Learning Theory
Statistical Learning
Face Detection
Object Detection
Object Recognition
Margin
Learning Algorithm
Theoretical Analysis
Robustness
Object detection
Vision
Narrative
Experimental Results
Experiment

All Science Journal Classification (ASJC) codes

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

Yang, M. H., Roth, D., & Ahuja, N. (2002). A tale of two classifiers: SNoW vs. SVM in visual recognition. In M. Nielsen, A. Heyden, G. Sparr, & P. Johansen (Eds.), Computer Vision - ECCV 2002 - 7th European Conference on Computer Vision, Proceedings (pp. 685-699). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 2353). Springer Verlag.
Yang, Ming Hsuan ; Roth, Dan ; Ahuja, Narendra. / A tale of two classifiers : SNoW vs. SVM in visual recognition. Computer Vision - ECCV 2002 - 7th European Conference on Computer Vision, Proceedings. editor / Mads Nielsen ; Anders Heyden ; Gunnar Sparr ; Peter Johansen. Springer Verlag, 2002. pp. 685-699 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{a402aadb7bce4531a602ba740b22c008,
title = "A tale of two classifiers: SNoW vs. SVM in visual recognition",
abstract = "Numerous statistical learning methods have been developed for visual recognition tasks. Few attempts, however, have been made to address theoretical issues, and in particular, study the suitability of different learning algorithms for visual recognition. Large margin classifiers, such as SNoW and SVM, have recently demonstrated their success in object detection and recognition. In this paper, we present a theoretical account of these two learning approaches, and their suitability to visual recognition. Using tools from computational learning theory, we show that the main difference between the generalization bounds of SVM and SNoW depends on the properties of the data. We argue that learning problems in the visual domain have sparseness characteristics and exhibit them by analyzing data taken from face detection experiments. Experimental results exhibit good generalization and robustness properties of the SNoW-based method, and conform to the theoretical analysis.",
author = "Yang, {Ming Hsuan} and Dan Roth and Narendra Ahuja",
year = "2002",
month = "1",
day = "1",
language = "English",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "685--699",
editor = "Mads Nielsen and Anders Heyden and Gunnar Sparr and Peter Johansen",
booktitle = "Computer Vision - ECCV 2002 - 7th European Conference on Computer Vision, Proceedings",
address = "Germany",

}

Yang, MH, Roth, D & Ahuja, N 2002, A tale of two classifiers: SNoW vs. SVM in visual recognition. in M Nielsen, A Heyden, G Sparr & P Johansen (eds), Computer Vision - ECCV 2002 - 7th European Conference on Computer Vision, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 2353, Springer Verlag, pp. 685-699, 7th European Conference on Computer Vision, ECCV 2002, Copenhagen, Denmark, 02/5/28.

A tale of two classifiers : SNoW vs. SVM in visual recognition. / Yang, Ming Hsuan; Roth, Dan; Ahuja, Narendra.

Computer Vision - ECCV 2002 - 7th European Conference on Computer Vision, Proceedings. ed. / Mads Nielsen; Anders Heyden; Gunnar Sparr; Peter Johansen. Springer Verlag, 2002. p. 685-699 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 2353).

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

TY - GEN

T1 - A tale of two classifiers

T2 - SNoW vs. SVM in visual recognition

AU - Yang, Ming Hsuan

AU - Roth, Dan

AU - Ahuja, Narendra

PY - 2002/1/1

Y1 - 2002/1/1

N2 - Numerous statistical learning methods have been developed for visual recognition tasks. Few attempts, however, have been made to address theoretical issues, and in particular, study the suitability of different learning algorithms for visual recognition. Large margin classifiers, such as SNoW and SVM, have recently demonstrated their success in object detection and recognition. In this paper, we present a theoretical account of these two learning approaches, and their suitability to visual recognition. Using tools from computational learning theory, we show that the main difference between the generalization bounds of SVM and SNoW depends on the properties of the data. We argue that learning problems in the visual domain have sparseness characteristics and exhibit them by analyzing data taken from face detection experiments. Experimental results exhibit good generalization and robustness properties of the SNoW-based method, and conform to the theoretical analysis.

AB - Numerous statistical learning methods have been developed for visual recognition tasks. Few attempts, however, have been made to address theoretical issues, and in particular, study the suitability of different learning algorithms for visual recognition. Large margin classifiers, such as SNoW and SVM, have recently demonstrated their success in object detection and recognition. In this paper, we present a theoretical account of these two learning approaches, and their suitability to visual recognition. Using tools from computational learning theory, we show that the main difference between the generalization bounds of SVM and SNoW depends on the properties of the data. We argue that learning problems in the visual domain have sparseness characteristics and exhibit them by analyzing data taken from face detection experiments. Experimental results exhibit good generalization and robustness properties of the SNoW-based method, and conform to the theoretical analysis.

UR - http://www.scopus.com/inward/record.url?scp=84937539670&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84937539670&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:84937539670

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 685

EP - 699

BT - Computer Vision - ECCV 2002 - 7th European Conference on Computer Vision, Proceedings

A2 - Nielsen, Mads

A2 - Heyden, Anders

A2 - Sparr, Gunnar

A2 - Johansen, Peter

PB - Springer Verlag

ER -

Yang MH, Roth D, Ahuja N. A tale of two classifiers: SNoW vs. SVM in visual recognition. In Nielsen M, Heyden A, Sparr G, Johansen P, editors, Computer Vision - ECCV 2002 - 7th European Conference on Computer Vision, Proceedings. Springer Verlag. 2002. p. 685-699. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).