With a Gaussian kernel function, we find that the distance between two classes (DBTC) can be used as a class separability criterion in feature space since the between-class separation and the within-class data distribution are taken into account impliedly. To test the validity of DBTC, we develop a method of tuning the kernel parameters in support vector machine (SVM) algorithm by maximizing the DBTC in feature space. Experimental results on the real-world data show that the proposed method consistently outperforms corresponding hyperparameters tuning methods.
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Copy
Jiancheng SUN, Chongxun ZHENG, Xiaohe LI, "Distance between Two Classes: A Novel Kernel Class Separability Criterion" in IEICE TRANSACTIONS on Information,
vol. E92-D, no. 7, pp. 1397-1400, July 2009, doi: 10.1587/transinf.E92.D.1397.
Abstract: With a Gaussian kernel function, we find that the distance between two classes (DBTC) can be used as a class separability criterion in feature space since the between-class separation and the within-class data distribution are taken into account impliedly. To test the validity of DBTC, we develop a method of tuning the kernel parameters in support vector machine (SVM) algorithm by maximizing the DBTC in feature space. Experimental results on the real-world data show that the proposed method consistently outperforms corresponding hyperparameters tuning methods.
URL: https://global.ieice.org/en_transactions/information/10.1587/transinf.E92.D.1397/_p
Copy
@ARTICLE{e92-d_7_1397,
author={Jiancheng SUN, Chongxun ZHENG, Xiaohe LI, },
journal={IEICE TRANSACTIONS on Information},
title={Distance between Two Classes: A Novel Kernel Class Separability Criterion},
year={2009},
volume={E92-D},
number={7},
pages={1397-1400},
abstract={With a Gaussian kernel function, we find that the distance between two classes (DBTC) can be used as a class separability criterion in feature space since the between-class separation and the within-class data distribution are taken into account impliedly. To test the validity of DBTC, we develop a method of tuning the kernel parameters in support vector machine (SVM) algorithm by maximizing the DBTC in feature space. Experimental results on the real-world data show that the proposed method consistently outperforms corresponding hyperparameters tuning methods.},
keywords={},
doi={10.1587/transinf.E92.D.1397},
ISSN={1745-1361},
month={July},}
Copy
TY - JOUR
TI - Distance between Two Classes: A Novel Kernel Class Separability Criterion
T2 - IEICE TRANSACTIONS on Information
SP - 1397
EP - 1400
AU - Jiancheng SUN
AU - Chongxun ZHENG
AU - Xiaohe LI
PY - 2009
DO - 10.1587/transinf.E92.D.1397
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E92-D
IS - 7
JA - IEICE TRANSACTIONS on Information
Y1 - July 2009
AB - With a Gaussian kernel function, we find that the distance between two classes (DBTC) can be used as a class separability criterion in feature space since the between-class separation and the within-class data distribution are taken into account impliedly. To test the validity of DBTC, we develop a method of tuning the kernel parameters in support vector machine (SVM) algorithm by maximizing the DBTC in feature space. Experimental results on the real-world data show that the proposed method consistently outperforms corresponding hyperparameters tuning methods.
ER -