Solução de problemas de otimização linear por redes neurais associadas a metodos de pontos interiores

AUTOR(ES)
DATA DE PUBLICAÇÃO

2003

RESUMO

This works explores possibilities of cooperation between neural networks and interior point methods to solve linear optimization problems. It seems that the neural networks and optimization communities carry on their research in worlds apart, with only tiny links between each other. Researchers in neural networks provided theoretical results for addressing optimization problems but did not go much beyond demonstrative examples; problems used to illustrate the optimization approaches by neural networks are small (most of them are textbook examples that can be solved by hand calculations). In this work, Hopfield neural networks and interior point methods are used in an integrated way to solve linear optimization problems. Hopfield networks perform the early stages of the optimization procedures, giving enhanced feasible starting points for interior point methods, which can be way ahead in the path to optimality. Cooperation with both the affine scale and primal-dual family of interior point methods is investigated. The approaches were applied to a set of real world linear programming problems, with different levels of guidance from the neural networks. The integrated approaches provide promising results, indicating that there may be a place for neural networks in solving large optimization problems.

ASSUNTO(S)

otimização matematica redes neurais (computação) programação linear

ACESSO AO ARTIGO

Documentos Relacionados