DocumentCode :
2495986
Title :
A nonexclusive task decomposition method for modular neural networks
Author :
Alves, Victor M O ; Cavalcanti, George D C
Author_Institution :
Center of Inf., Fed. Univ. of Pernambuco, Recife, Brazil
fYear :
2010
fDate :
18-23 July 2010
Firstpage :
1
Lastpage :
6
Abstract :
Modular neural networks (MNNs) architectures have been developed aiming to outperform single neural nets. One of the main drawbacks in the construction of the MNNs is the task decomposition which consists in divide the problem into simpler sub-problems. This paper proposes a novel task decomposition method in which the classes of the problem can be divided redundantly. Thus, two different expert modules can have the same class. This is specially interesting for problems that have multimodal classes. The proposed MNN, called Redundant Pattern Distributor, is compared with other ones over many databases and the results show its effectiveness.
Keywords :
neural nets; modular neural networks; nonexclusive task decomposition method; redundant pattern distributor; single neural nets; Artificial neural networks; Computer architecture; Databases; Image segmentation; Neurons; Satellites; Training;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Neural Networks (IJCNN), The 2010 International Joint Conference on
Conference_Location :
Barcelona
ISSN :
1098-7576
Print_ISBN :
978-1-4244-6916-1
Type :
conf
DOI :
10.1109/IJCNN.2010.5596840
Filename :
5596840
Link To Document :
بازگشت