Analise e sintese de estrategias de aprendizado para redes neurais artificiais

AUTOR(ES)

Leandro Nunes de Castro Silva

DATA DE PUBLICAÇÃO

1998

RESUMO

This dissertation aims at developing, analysing and implementing training strategies for neural networks, involving mainly methods that take into account the second order information of feedforward multilayer perceptron networks. As a complement to the study of multilayer architectures, a detailed analysis of a self-organized model is performed, and some alternatives for determining optimal dimensionality of the architecture and efficient initial conditions are also proposed. The results of this work are supposed to contribute in the sense of annoucing to the scientific community that the artificial neural networks field is conc1uding a long and necessary transition between an initial phase, characterized by a great deal of heuristic content, and a phase of scientific maturity, characterized by the theoretical formalization of the results and strong interaction with other well stablished scientific fields

ASSUNTO(S)

inteligencia artificial programação não-linear redes neurais (computação)

Documentos Relacionados

• Redes neurais artificiais e análise de regressão para estimativa de volume de espécies nativas
• Um modelo estratégico para a análise de crédito utilizando redes neurais artificiais
• Treinamento de redes neurais artificiais baseado em sistemas de estrutura variável com taxa de aprendizado adaptativa
• Utilização de redes neurais artificiais para detecção e diagnóstico de falhas
• Redes neurais artificiais para estimação do volume de árvores