Using ANN back-propagation technique to represent the group of ILD patterns
Balemir Uragun · 2017
This study aims for the clustered data representation from recognized-patterns using the feedforward neural networks (ANN); a group of clustered data can be typified with a group of recognized patterns or called master template. This master template consists of the different numbers of clustered data and to be used by a suitable ANN for the clustered data representation. It is an objective fact that through a standard ANN processes of (a) data-oriented parameter selection, (b) identification of the appropriate number of processing layers with the best performance, learning method and training algorithms, and finally (c) development of the clustered data representation. Initially, a Supervised-Feedforward ANN was selected from the literature, based on applications similar to those in this study. Here, the appropriately optimized ANN architecture was tested with a variety of types of training algorithms, and the most useful training algorithm for this specific application was determined.