DocumentCode :
2328862
Title :
Adaptive task decomposition and modular multilayer perceptrons for letter recognition
Author :
Daqi, Gao ; Renliang, Li ; Guiping, Nie ; Changwu, Li
Author_Institution :
Dept. of Comput., East China Univ. of Sci. & Technol., Shanghai, China
Volume :
4
fYear :
2004
fDate :
25-29 July 2004
Firstpage :
2937
Abstract :
This paper proposes a task decomposition method, which divides a large-scale learning problem into multiple limited-scale pairs of training subsets and cross validation (CV) subsets. Correspondingly, modular multilayer perceptrons are set up. At first, one training subset only consists of its own class and several most neighboring categories, and then some classes in the CV subset are moved into it according to the generalization error of the module. This work presents an empirical formula for selecting the initial number of hidden nodes, and a method for determining the optimal number of hidden units with the help of singular value decomposition. The result for letter recognition shows that the above methods are quite effective.
Keywords :
character recognition; learning (artificial intelligence); multilayer perceptrons; set theory; singular value decomposition; adaptive task decomposition method; cross validation subsets; large-scale learning problem; letter recognition; modular multilayer perceptrons; singular value decomposition; training subsets; Bioreactors; Independent component analysis; Laboratories; Large-scale systems; Multilayer perceptrons; Nonhomogeneous media; Paper technology; Pattern recognition; Support vector machine classification; Support vector machines;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Neural Networks, 2004. Proceedings. 2004 IEEE International Joint Conference on
ISSN :
1098-7576
Print_ISBN :
0-7803-8359-1
Type :
conf
DOI :
10.1109/IJCNN.2004.1381131
Filename :
1381131
Link To Document :
بازگشت