On initialization of back propagation neural networks for land cover classification using satellite imageries

Author

Barros, Maria Suelena Santiago ; Nowosad, Alexandre Guirland ; De Andrade Amorin, M.C.

Author_Institution

Inst. Nacional de Pesquisas Espaciais, Sao Jose dos Campos, Brazil

fYear

1993

fDate

18-21 Aug 1993

Firstpage

902

Abstract

The use of feedback in back propagation neural networks satellite imagery land cover classification is investigated. Two networks are used and the described method can be considered as a discrete-time stochastic evolutionary process. The input to the process is an imagery pair plus the set of all weights of the last step´s final configuration. The final answer would be a subset of weights corresponding to one of the networks

Keywords

backpropagation; geophysical techniques; image recognition; neural nets; remote sensing; back propagation neural network; discrete-time stochastic evolutionary process; feedback; geophysical measurement technique; image classification; imagery pair; initialization; land cover; land surface imaging; landcover classification; neural net; remote sensing; satellite imagery; terrain mapping; vegetation; Computational modeling; Image converters; Neural networks; Neurofeedback; Neurons; Random sequences; Random variables; Satellites; Stochastic processes;

fLanguage

English

Publisher

ieee

Conference_Titel

Geoscience and Remote Sensing Symposium, 1993. IGARSS '93. Better Understanding of Earth Environment., International

Conference_Location

Tokyo

Print_ISBN

0-7803-1240-6

Type

conf

DOI

10.1109/IGARSS.1993.322193

Filename

322193