Application of a Modular Feedforward Neural Network for Grade Estimation

摘要: This article presents new neural network (NN) architecture to improve its ability for grade estimation. The main aim of this study is use a specific NN which has simpler and consequently achieve better solution. Most the commonly used NNs have fully established connection among their nodes, necessitates multivariable objective function be optimized. Therefore, more number variables in function, complexity NN. leads trap local minima. In study, NN, connections based on final performance are eliminated, used. Toward aim, several architectures were tested, finally yielded minimum error was selected. selected low nodes help learning algorithm converge rapidly accurately. Furthermore, deal with small data sets. For testing evaluating method, case an iron deposit performed. Also, compare obtained results, some common techniques estimation, e.g., geostatistics multilayer perceptron (MLP) According shows