DocumentCode :
2481447
Title :
A Unifying Framework for Learning the Linear Combiners for Classifier Ensembles
Author :
Erdogan, Hakan ; Sen, Mehmet Umut
Author_Institution :
Fac. of Eng. & Natural Sci., Sabanci Univ., Istanbul, Turkey
fYear :
2010
fDate :
23-26 Aug. 2010
Firstpage :
2985
Lastpage :
2988
Abstract :
For classifier ensembles, an effective combination method is to combine the outputs of each classifier using a linearly weighted combination rule. There are multiple ways to linearly combine classifier outputs and it is beneficial to analyze them as a whole. We present a unifying framework for multiple linear combination types in this paper. This unification enables using the same learning algorithms for different types of linear combiners. We present various ways to train the weights using regularized empirical loss minimization. We propose using the hinge loss for better performance as compared to the conventional least-squares loss. We analyze the effects of using hinge loss for various types of linear weight training by running experiments on three different databases. We show that, in certain problems, linear combiners with fewer parameters may perform as well as the ones with much larger number of parameters even in the presence of regularization.
Keywords :
learning (artificial intelligence); least squares approximations; minimisation; pattern classification; classifier ensemble; conventional least squares loss; linear combiner learning; linear weight training; linearly combine classifier; linearly weighted combination rule; multiple linear combination; regularized empirical loss minimization; unifying framework; Accuracy; Databases; Fasteners; Minimization; Training; Training data; Vectors; classifier fusion; linear classifier learning; linear combiners; stacked generalization;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Pattern Recognition (ICPR), 2010 20th International Conference on
Conference_Location :
Istanbul
ISSN :
1051-4651
Print_ISBN :
978-1-4244-7542-1
Type :
conf
DOI :
10.1109/ICPR.2010.731
Filename :
5595979
Link To Document :
بازگشت