Best feature selection using successive elimination of poor performers

K. J. Siddiqui, E. C. Greco, N. Kadri, Syed M. Mohiuddin, M. Sketch

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

Abstract

This paper addresses the issue of feature extraction and selection, focusing particularly ion the feature selection issue. Without assuming any particular classification algorithm it suggests that first one should extract as much information (features) as conveniently possible and then apply the proposed successive elimination process to remove redundant and poor features and then select a significantly smaller, yet useful, feature subset that enhances the performance of the classifier. The algorithm is formally described and is successfully applied to a four class ECG classification problem. A minimum distance classifier (MDC) using Mahalanobis distance as the decision criterion is developed. Using MDC an overall recognition performance of 87.5% is obtained on the testing set of the four ECG classes.

Original languageEnglish
Title of host publicationProceedings of the Annual Conference on Engineering in Medicine and Biology
EditorsAndrew Y.J. Szeto, Rangaraj M. Rangayyan
PublisherPubl by IEEE
Pages725-726
Number of pages2
Volume15
Editionpt 2
ISBN (Print)0780313771
StatePublished - 1993
EventProceedings of the 15th Annual International Conference of the IEEE Engineering in Medicine and Biology Society. Part 2 (of 3) - San Diego, CA, USA
Duration: Oct 28 1993Oct 31 1993

Other

OtherProceedings of the 15th Annual International Conference of the IEEE Engineering in Medicine and Biology Society. Part 2 (of 3)
CitySan Diego, CA, USA
Period10/28/9310/31/93

Fingerprint

Feature extraction
Classifiers
Electrocardiography
Ions
Testing

All Science Journal Classification (ASJC) codes

  • Bioengineering

Cite this

Siddiqui, K. J., Greco, E. C., Kadri, N., Mohiuddin, S. M., & Sketch, M. (1993). Best feature selection using successive elimination of poor performers. In A. Y. J. Szeto, & R. M. Rangayyan (Eds.), Proceedings of the Annual Conference on Engineering in Medicine and Biology (pt 2 ed., Vol. 15, pp. 725-726). Publ by IEEE.

Best feature selection using successive elimination of poor performers. / Siddiqui, K. J.; Greco, E. C.; Kadri, N.; Mohiuddin, Syed M.; Sketch, M.

Proceedings of the Annual Conference on Engineering in Medicine and Biology. ed. / Andrew Y.J. Szeto; Rangaraj M. Rangayyan. Vol. 15 pt 2. ed. Publ by IEEE, 1993. p. 725-726.

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

Siddiqui, KJ, Greco, EC, Kadri, N, Mohiuddin, SM & Sketch, M 1993, Best feature selection using successive elimination of poor performers. in AYJ Szeto & RM Rangayyan (eds), Proceedings of the Annual Conference on Engineering in Medicine and Biology. pt 2 edn, vol. 15, Publ by IEEE, pp. 725-726, Proceedings of the 15th Annual International Conference of the IEEE Engineering in Medicine and Biology Society. Part 2 (of 3), San Diego, CA, USA, 10/28/93.
Siddiqui KJ, Greco EC, Kadri N, Mohiuddin SM, Sketch M. Best feature selection using successive elimination of poor performers. In Szeto AYJ, Rangayyan RM, editors, Proceedings of the Annual Conference on Engineering in Medicine and Biology. pt 2 ed. Vol. 15. Publ by IEEE. 1993. p. 725-726
Siddiqui, K. J. ; Greco, E. C. ; Kadri, N. ; Mohiuddin, Syed M. ; Sketch, M. / Best feature selection using successive elimination of poor performers. Proceedings of the Annual Conference on Engineering in Medicine and Biology. editor / Andrew Y.J. Szeto ; Rangaraj M. Rangayyan. Vol. 15 pt 2. ed. Publ by IEEE, 1993. pp. 725-726
@inproceedings{06612b8097c546d5ad3b2f8459ab3514,
title = "Best feature selection using successive elimination of poor performers",
abstract = "This paper addresses the issue of feature extraction and selection, focusing particularly ion the feature selection issue. Without assuming any particular classification algorithm it suggests that first one should extract as much information (features) as conveniently possible and then apply the proposed successive elimination process to remove redundant and poor features and then select a significantly smaller, yet useful, feature subset that enhances the performance of the classifier. The algorithm is formally described and is successfully applied to a four class ECG classification problem. A minimum distance classifier (MDC) using Mahalanobis distance as the decision criterion is developed. Using MDC an overall recognition performance of 87.5{\%} is obtained on the testing set of the four ECG classes.",
author = "Siddiqui, {K. J.} and Greco, {E. C.} and N. Kadri and Mohiuddin, {Syed M.} and M. Sketch",
year = "1993",
language = "English",
isbn = "0780313771",
volume = "15",
pages = "725--726",
editor = "Szeto, {Andrew Y.J.} and Rangayyan, {Rangaraj M.}",
booktitle = "Proceedings of the Annual Conference on Engineering in Medicine and Biology",
publisher = "Publ by IEEE",
edition = "pt 2",

}

TY - GEN

T1 - Best feature selection using successive elimination of poor performers

AU - Siddiqui, K. J.

AU - Greco, E. C.

AU - Kadri, N.

AU - Mohiuddin, Syed M.

AU - Sketch, M.

PY - 1993

Y1 - 1993

N2 - This paper addresses the issue of feature extraction and selection, focusing particularly ion the feature selection issue. Without assuming any particular classification algorithm it suggests that first one should extract as much information (features) as conveniently possible and then apply the proposed successive elimination process to remove redundant and poor features and then select a significantly smaller, yet useful, feature subset that enhances the performance of the classifier. The algorithm is formally described and is successfully applied to a four class ECG classification problem. A minimum distance classifier (MDC) using Mahalanobis distance as the decision criterion is developed. Using MDC an overall recognition performance of 87.5% is obtained on the testing set of the four ECG classes.

AB - This paper addresses the issue of feature extraction and selection, focusing particularly ion the feature selection issue. Without assuming any particular classification algorithm it suggests that first one should extract as much information (features) as conveniently possible and then apply the proposed successive elimination process to remove redundant and poor features and then select a significantly smaller, yet useful, feature subset that enhances the performance of the classifier. The algorithm is formally described and is successfully applied to a four class ECG classification problem. A minimum distance classifier (MDC) using Mahalanobis distance as the decision criterion is developed. Using MDC an overall recognition performance of 87.5% is obtained on the testing set of the four ECG classes.

UR - http://www.scopus.com/inward/record.url?scp=0027842044&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0027842044&partnerID=8YFLogxK

M3 - Conference contribution

SN - 0780313771

VL - 15

SP - 725

EP - 726

BT - Proceedings of the Annual Conference on Engineering in Medicine and Biology

A2 - Szeto, Andrew Y.J.

A2 - Rangayyan, Rangaraj M.

PB - Publ by IEEE

ER -