DocumentCode
305385
Title
A stochastic approach to layered neural networks
Author
Pedreira, C.E. ; Parente, E.
Author_Institution
Catholic Univ. of Rio de Janeiro, Brazil
Volume
3
fYear
1996
fDate
14-17 Oct 1996
Firstpage
1947
Abstract
We propose a stochastic framework to backpropagation. The network inputs are random variables, and thus a stochastic output is produced. The application range of this new approach goes from medical to financial problems. In this paper, a classification method for a two-group discriminant problem based on the interval algebra is proposed. This approach is limited to classification problems, and can not be directly extended to more than two groups
Keywords
backpropagation; decision support systems; feedforward neural nets; medical diagnostic computing; pattern classification; backpropagation; decision support system; feedforward neural networks; interval algebra; medical diagnostic system; multilayer neural networks; pattern classification; stochastic neural networks; stochastic output; two-group discriminant problem; Algebra; Backpropagation; Cost function; Feedforward neural networks; Joining processes; Medical diagnostic imaging; Neural networks; Random variables; Stochastic processes; Uncertainty;
fLanguage
English
Publisher
ieee
Conference_Titel
Systems, Man, and Cybernetics, 1996., IEEE International Conference on
Conference_Location
Beijing
ISSN
1062-922X
Print_ISBN
0-7803-3280-6
Type
conf
DOI
10.1109/ICSMC.1996.565419
Filename
565419
Link To Document