Improved TV-division output coding for multiclass learning problems

Jaepil Ko, Eunju Kim, Hyeran Byun

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

2 Citations (Scopus)

Abstract

The output coding for multiclass learning problems is a generalization of one-per-class, all-pairs, and error correcting output codes. Although, the prevailing concepts of output coding has been error correcting properties, the one-per-class and all-pairs are still considered to be one of the state-of-art methods. However, these two methods are contrary to each other in the aspect of producing complex dichotomies and the problem of nonsense outputs. In additions, they all perform a prior decomposition without regards to the properties of a given training data set. In this paper, we propose a new data-driven output coding method that is the generalized form of one-per-class and all-pairs. We present the properties of the proposed method. From experimental results on both a toy problem and real benchmark datasets, we present that our proposed method achieves a comparable performance with good properties.

Original languageEnglish
Title of host publicationProceedings of the 17th International Conference on Pattern Recognition, ICPR 2004
EditorsJ. Kittler, M. Petrou, M. Nixon
Pages470-473
Number of pages4
DOIs
Publication statusPublished - 2004 Dec 20
EventProceedings of the 17th International Conference on Pattern Recognition, ICPR 2004 - Cambridge, United Kingdom
Duration: 2004 Aug 232004 Aug 26

Publication series

NameProceedings - International Conference on Pattern Recognition
Volume3
ISSN (Print)1051-4651

Other

OtherProceedings of the 17th International Conference on Pattern Recognition, ICPR 2004
CountryUnited Kingdom
CityCambridge
Period04/8/2304/8/26

Fingerprint

Decomposition

All Science Journal Classification (ASJC) codes

  • Computer Vision and Pattern Recognition

Cite this

Ko, J., Kim, E., & Byun, H. (2004). Improved TV-division output coding for multiclass learning problems. In J. Kittler, M. Petrou, & M. Nixon (Eds.), Proceedings of the 17th International Conference on Pattern Recognition, ICPR 2004 (pp. 470-473). (Proceedings - International Conference on Pattern Recognition; Vol. 3). https://doi.org/10.1109/ICPR.2004.1334568
Ko, Jaepil ; Kim, Eunju ; Byun, Hyeran. / Improved TV-division output coding for multiclass learning problems. Proceedings of the 17th International Conference on Pattern Recognition, ICPR 2004. editor / J. Kittler ; M. Petrou ; M. Nixon. 2004. pp. 470-473 (Proceedings - International Conference on Pattern Recognition).
@inproceedings{888490f78999450bb5f4fe966d68737c,
title = "Improved TV-division output coding for multiclass learning problems",
abstract = "The output coding for multiclass learning problems is a generalization of one-per-class, all-pairs, and error correcting output codes. Although, the prevailing concepts of output coding has been error correcting properties, the one-per-class and all-pairs are still considered to be one of the state-of-art methods. However, these two methods are contrary to each other in the aspect of producing complex dichotomies and the problem of nonsense outputs. In additions, they all perform a prior decomposition without regards to the properties of a given training data set. In this paper, we propose a new data-driven output coding method that is the generalized form of one-per-class and all-pairs. We present the properties of the proposed method. From experimental results on both a toy problem and real benchmark datasets, we present that our proposed method achieves a comparable performance with good properties.",
author = "Jaepil Ko and Eunju Kim and Hyeran Byun",
year = "2004",
month = "12",
day = "20",
doi = "10.1109/ICPR.2004.1334568",
language = "English",
isbn = "0769521282",
series = "Proceedings - International Conference on Pattern Recognition",
pages = "470--473",
editor = "J. Kittler and M. Petrou and M. Nixon",
booktitle = "Proceedings of the 17th International Conference on Pattern Recognition, ICPR 2004",

}

Ko, J, Kim, E & Byun, H 2004, Improved TV-division output coding for multiclass learning problems. in J Kittler, M Petrou & M Nixon (eds), Proceedings of the 17th International Conference on Pattern Recognition, ICPR 2004. Proceedings - International Conference on Pattern Recognition, vol. 3, pp. 470-473, Proceedings of the 17th International Conference on Pattern Recognition, ICPR 2004, Cambridge, United Kingdom, 04/8/23. https://doi.org/10.1109/ICPR.2004.1334568

Improved TV-division output coding for multiclass learning problems. / Ko, Jaepil; Kim, Eunju; Byun, Hyeran.

Proceedings of the 17th International Conference on Pattern Recognition, ICPR 2004. ed. / J. Kittler; M. Petrou; M. Nixon. 2004. p. 470-473 (Proceedings - International Conference on Pattern Recognition; Vol. 3).

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

TY - GEN

T1 - Improved TV-division output coding for multiclass learning problems

AU - Ko, Jaepil

AU - Kim, Eunju

AU - Byun, Hyeran

PY - 2004/12/20

Y1 - 2004/12/20

N2 - The output coding for multiclass learning problems is a generalization of one-per-class, all-pairs, and error correcting output codes. Although, the prevailing concepts of output coding has been error correcting properties, the one-per-class and all-pairs are still considered to be one of the state-of-art methods. However, these two methods are contrary to each other in the aspect of producing complex dichotomies and the problem of nonsense outputs. In additions, they all perform a prior decomposition without regards to the properties of a given training data set. In this paper, we propose a new data-driven output coding method that is the generalized form of one-per-class and all-pairs. We present the properties of the proposed method. From experimental results on both a toy problem and real benchmark datasets, we present that our proposed method achieves a comparable performance with good properties.

AB - The output coding for multiclass learning problems is a generalization of one-per-class, all-pairs, and error correcting output codes. Although, the prevailing concepts of output coding has been error correcting properties, the one-per-class and all-pairs are still considered to be one of the state-of-art methods. However, these two methods are contrary to each other in the aspect of producing complex dichotomies and the problem of nonsense outputs. In additions, they all perform a prior decomposition without regards to the properties of a given training data set. In this paper, we propose a new data-driven output coding method that is the generalized form of one-per-class and all-pairs. We present the properties of the proposed method. From experimental results on both a toy problem and real benchmark datasets, we present that our proposed method achieves a comparable performance with good properties.

UR - http://www.scopus.com/inward/record.url?scp=10044252059&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=10044252059&partnerID=8YFLogxK

U2 - 10.1109/ICPR.2004.1334568

DO - 10.1109/ICPR.2004.1334568

M3 - Conference contribution

AN - SCOPUS:10044252059

SN - 0769521282

T3 - Proceedings - International Conference on Pattern Recognition

SP - 470

EP - 473

BT - Proceedings of the 17th International Conference on Pattern Recognition, ICPR 2004

A2 - Kittler, J.

A2 - Petrou, M.

A2 - Nixon, M.

ER -

Ko J, Kim E, Byun H. Improved TV-division output coding for multiclass learning problems. In Kittler J, Petrou M, Nixon M, editors, Proceedings of the 17th International Conference on Pattern Recognition, ICPR 2004. 2004. p. 470-473. (Proceedings - International Conference on Pattern Recognition). https://doi.org/10.1109/ICPR.2004.1334568