Robust semi-supervised SVM on kernel partial least discriminant space for high dimensional data mining

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

Abstract

Kernel machines (such as support vector machines) have demonstrated excellent performance in numerous areas of pattern recognitions. However, traditional kernel machines do not make efficient use of both labeled training data and unlabeled testing data. Moreover, high dimensional and nonlinear distributed data generally degrade the performance of kernel classifiers due to the curse of dimensionality. To address these problems, this study proposes a novel hybrid classifier which constructs a robust semi- supervised support vector machine (SVM) on kernel partial least square discriminant space (KPLSDS). KPLSDS is created by optimal projection of original data space to a representative low dimensional subspace which has maximum covariance between inputs and outputs. Robust semi-supervised SVMs on KPLSDS exploit the candidate low-density separators and simultaneously prevent identifying a poor separator from the help of unlabeled data. Compared with other dimensionality reduction methods and conventional classifiers, the hybrid classifier performs best.

Original languageEnglish
Title of host publication2012 International Conference on Information Science and Applications, ICISA 2012
DOIs
Publication statusPublished - 2012 Jul 30
Event2012 International Conference on Information Science and Applications, ICISA 2012 - Suwon, Korea, Republic of
Duration: 2012 May 232012 May 25

Publication series

Name2012 International Conference on Information Science and Applications, ICISA 2012

Other

Other2012 International Conference on Information Science and Applications, ICISA 2012
CountryKorea, Republic of
CitySuwon
Period12-05-2312-05-25

Fingerprint

Support vector machines
Data mining
Classifiers
Separators
Pattern recognition
Testing

All Science Journal Classification (ASJC) codes

  • Computer Science Applications
  • Information Systems

Cite this

Huang, S-C., & Wu, T-K. (2012). Robust semi-supervised SVM on kernel partial least discriminant space for high dimensional data mining. In 2012 International Conference on Information Science and Applications, ICISA 2012 [6220924] (2012 International Conference on Information Science and Applications, ICISA 2012). https://doi.org/10.1109/ICISA.2012.6220924
Huang, Shian-Chang ; Wu, Tung-Kuang. / Robust semi-supervised SVM on kernel partial least discriminant space for high dimensional data mining. 2012 International Conference on Information Science and Applications, ICISA 2012. 2012. (2012 International Conference on Information Science and Applications, ICISA 2012).
@inproceedings{28e2bbe215ff4aa4a7710ba5969c4efe,
title = "Robust semi-supervised SVM on kernel partial least discriminant space for high dimensional data mining",
abstract = "Kernel machines (such as support vector machines) have demonstrated excellent performance in numerous areas of pattern recognitions. However, traditional kernel machines do not make efficient use of both labeled training data and unlabeled testing data. Moreover, high dimensional and nonlinear distributed data generally degrade the performance of kernel classifiers due to the curse of dimensionality. To address these problems, this study proposes a novel hybrid classifier which constructs a robust semi- supervised support vector machine (SVM) on kernel partial least square discriminant space (KPLSDS). KPLSDS is created by optimal projection of original data space to a representative low dimensional subspace which has maximum covariance between inputs and outputs. Robust semi-supervised SVMs on KPLSDS exploit the candidate low-density separators and simultaneously prevent identifying a poor separator from the help of unlabeled data. Compared with other dimensionality reduction methods and conventional classifiers, the hybrid classifier performs best.",
author = "Shian-Chang Huang and Tung-Kuang Wu",
year = "2012",
month = "7",
day = "30",
doi = "10.1109/ICISA.2012.6220924",
language = "English",
isbn = "9781467321860",
series = "2012 International Conference on Information Science and Applications, ICISA 2012",
booktitle = "2012 International Conference on Information Science and Applications, ICISA 2012",

}

Huang, S-C & Wu, T-K 2012, Robust semi-supervised SVM on kernel partial least discriminant space for high dimensional data mining. in 2012 International Conference on Information Science and Applications, ICISA 2012., 6220924, 2012 International Conference on Information Science and Applications, ICISA 2012, 2012 International Conference on Information Science and Applications, ICISA 2012, Suwon, Korea, Republic of, 12-05-23. https://doi.org/10.1109/ICISA.2012.6220924

Robust semi-supervised SVM on kernel partial least discriminant space for high dimensional data mining. / Huang, Shian-Chang; Wu, Tung-Kuang.

2012 International Conference on Information Science and Applications, ICISA 2012. 2012. 6220924 (2012 International Conference on Information Science and Applications, ICISA 2012).

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

TY - GEN

T1 - Robust semi-supervised SVM on kernel partial least discriminant space for high dimensional data mining

AU - Huang, Shian-Chang

AU - Wu, Tung-Kuang

PY - 2012/7/30

Y1 - 2012/7/30

N2 - Kernel machines (such as support vector machines) have demonstrated excellent performance in numerous areas of pattern recognitions. However, traditional kernel machines do not make efficient use of both labeled training data and unlabeled testing data. Moreover, high dimensional and nonlinear distributed data generally degrade the performance of kernel classifiers due to the curse of dimensionality. To address these problems, this study proposes a novel hybrid classifier which constructs a robust semi- supervised support vector machine (SVM) on kernel partial least square discriminant space (KPLSDS). KPLSDS is created by optimal projection of original data space to a representative low dimensional subspace which has maximum covariance between inputs and outputs. Robust semi-supervised SVMs on KPLSDS exploit the candidate low-density separators and simultaneously prevent identifying a poor separator from the help of unlabeled data. Compared with other dimensionality reduction methods and conventional classifiers, the hybrid classifier performs best.

AB - Kernel machines (such as support vector machines) have demonstrated excellent performance in numerous areas of pattern recognitions. However, traditional kernel machines do not make efficient use of both labeled training data and unlabeled testing data. Moreover, high dimensional and nonlinear distributed data generally degrade the performance of kernel classifiers due to the curse of dimensionality. To address these problems, this study proposes a novel hybrid classifier which constructs a robust semi- supervised support vector machine (SVM) on kernel partial least square discriminant space (KPLSDS). KPLSDS is created by optimal projection of original data space to a representative low dimensional subspace which has maximum covariance between inputs and outputs. Robust semi-supervised SVMs on KPLSDS exploit the candidate low-density separators and simultaneously prevent identifying a poor separator from the help of unlabeled data. Compared with other dimensionality reduction methods and conventional classifiers, the hybrid classifier performs best.

UR - http://www.scopus.com/inward/record.url?scp=84864232811&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84864232811&partnerID=8YFLogxK

U2 - 10.1109/ICISA.2012.6220924

DO - 10.1109/ICISA.2012.6220924

M3 - Conference contribution

AN - SCOPUS:84864232811

SN - 9781467321860

T3 - 2012 International Conference on Information Science and Applications, ICISA 2012

BT - 2012 International Conference on Information Science and Applications, ICISA 2012

ER -

Huang S-C, Wu T-K. Robust semi-supervised SVM on kernel partial least discriminant space for high dimensional data mining. In 2012 International Conference on Information Science and Applications, ICISA 2012. 2012. 6220924. (2012 International Conference on Information Science and Applications, ICISA 2012). https://doi.org/10.1109/ICISA.2012.6220924