Título: Avaliação Prática de Redes Neurais para o Acasalamento de Gado Nelore
Autores: Franco, Rômulo B.; Martins, Weber
Resumo: This paper presents a system designed to predict the best bull should be chosen in order to obtain the best result given information on a specific cow. Learning is conducted by using data from real work done by specialists on multilayer perceptron networks and variants of backpropagation algorithm. The main goal was to find the best single neural network the could perform the task with confidence. This study provided some interesting results on the efficiency of backpropagation standard, with momentum and employing the technique of weight decay. Many factors were varied, including number of cycles, learning rate, momentum rate, topology, etc.
Artigo em pdf: 4cbrn_053.pdf
Arquivo BibTex: 4cbrn_053.bib