Neuro-Dea: A New Optimization Algorithm

Título: Neuro-Dea: A New Optimization Algorithm

Autores: Biondi Neto, Luiz; Lins, Marcos Estellita; Chiganer, Luis; Fukuda, Fernando Hideo; Roberto Junior, Vincenzo De; Antonio, Elias Restum; Oliveira, Fabiano S.

Resumo: This paper investigates the use of Data Envelopment Analysis (DEA) combined with Artificial Neural Networks (ANN) to produce a new hybrid optimization structure called Neuro-DEA. The basic idea is build an unconventional ANN where the optimizations modules used in DEA analysis are implemented in high-speed optimization structure called Neuro-LP. The Neuro-LP structure represent an ANN who implements a numerical solution based on gradient method. So the linear programming problem (LPP) is converted into an optimization problem without constraints by using a pseudo-cost function and penalty techniques. The Neuro-LP structure represents each Decision-making unit (DMU) and may be connected assembling the final proposed Neuro-DEA structure.


Páginas: 6

Código DOI: 10.21528/CBRN2001-069

Artigo em pdf: 5cbrn_069.pdf

Arquivo BibTex: 5cbrn_069.bib