@InProceedings{VijaykumarStePreCamNow:2002:OpNeNe,
author = "Vijaykumar, Nandamudi Lankalapali and Stephanyl, Stephan and
Preto, Airam Jonatas and Campos Velho, Haroldo Fraga de and
Nowosad, Alexandre G.",
affiliation = "{INPE-Sao Jose dos Campos-12227-010-SP-Brasil}",
title = "Optimized Neural Network Code for Data Assimilation",
booktitle = "Anais...",
year = "2002",
pages = "3841--3849",
organization = "Congresso Brasileiro de Meteorologia, 12.",
abstract = "The data assimilation process can be described as a procedure that
uses observational data to improve the prediction made by an
inaccurate mathematical modelo Recent1y, neural networks have been
proposed as a new method for data assimilation. The Multilayer
Perceptron network with backpropagation learning was chosen for
this procedure. Neural networks are inherent1y a parallel
procedure. This paper presents some strategies being used to
achieve an optimized parallel code for the network training. Code
optimizations include the use of either High Perfonnance Fortran
directives or Message Passing Interface library calls. A neural
network for Data Assimilation was trained based on both the
physical models of the Lorenz and shallow water equations.",
conference-location = "Foz do Iguacu",
conference-year = "4-9 ago.",
language = "en",
organisation = "SBMET",
urlaccessdate = "29 jun. 2024"
}