DocumentCode :
2934463
Title :
Boosting Kernel Discriminant Analysis for pattern classification
Author :
Kita, Shinji ; Ozawa, Seiichi ; Maekawa, Satoshi ; Abe, Shigeo
Author_Institution :
Kobe Univ., Kobe
fYear :
2007
fDate :
Nov. 28 2007-Dec. 1 2007
Firstpage :
658
Lastpage :
661
Abstract :
This paper presents a new boosting algorithm called Boosting Kernel Discriminant Analysis (BKDA) in which the feature selection and the classifier training are conducted by Kernel Discriminant Analysis (KDA) and AdaBoost.M2, respectively. To reduce the dependency between classifier outputs and to speed up the learning, each classifier is trained in the different feature space which is obtained by applying KDA to a small set of hard-to-classify training samples. The proposed BKDA is evaluated using standard benchmark datasets. The experimental results demonstrate that BKDA outperforms both Boosting Linear Discriminant Analysis (BLDA) and Support Vector Machine (S VM) for multi-class classification problems. On the other hand, the performance evaluation for 2-class problems shows that the advantage of the proposed BKDA against BLDA and SVM depends on the datasets.
Keywords :
feature extraction; learning (artificial intelligence); pattern classification; boosting kernel discriminant analysis; classifier training; feature selection; pattern classification; Algorithm design and analysis; Boosting; Information analysis; Kernel; Linear discriminant analysis; Pattern analysis; Pattern classification; Signal analysis; Support vector machine classification; Support vector machines; Boosting; Feature Selection; Kernel Discriminant Analysis; Neural Networks; Pattern Classification;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Intelligent Signal Processing and Communication Systems, 2007. ISPACS 2007. International Symposium on
Conference_Location :
Xiamen
Print_ISBN :
978-1-4244-1447-5
Electronic_ISBN :
978-1-4244-1447-5
Type :
conf
DOI :
10.1109/ISPACS.2007.4445973
Filename :
4445973
Link To Document :
بازگشت