Naturally intelligent systems
Introduction to the theory of neural computation
Introduction to the theory of neural computation
Introduction to artificial neural systems
Introduction to artificial neural systems
Practical neural network recipes in C++
Practical neural network recipes in C++
Advanced algorithms for neural networks: a C++ sourcebook
Advanced algorithms for neural networks: a C++ sourcebook
Understanding Neural Networks; Computer Explorations
Understanding Neural Networks; Computer Explorations
ICCIMA '99 Proceedings of the 3rd International Conference on Computational Intelligence and Multimedia Applications
| Hi-index | 0.00 |
While there are many models of the neural networks that are suitable for a particular application, each model will yield different accuracy when applied and moreover, the features of the training data that are used by the neural networks will differ in each instance. When a neural network is initially trained for a specific application, some of the features of the training data are not significant to the network's decisions, while other features are critical. Further, there is a cost and difficulty of measurement associated with the collection of process parameter data to be used as inputs in the neural network model. Hence, from industry point of view, it is beneficial to include only value-adding parameters in the neural network model, to avoid expenditure in collecting irrelevant process data that can be omitted without compromising the model accuracy. In this paper, a technique is used to identify the significant process parameters that are required for a particular application. Using a practical application in smelting industry, non-contributing variables are removed from the neural network model to achieve an improvement in prediction accuracy.