APPLICATION OF A FEEDFORWARD ARTIFICIAL NEURAL-NETWORK AS A MAPPING DEVICE

The article presents the ability to use a feed-forward neural network as a mapping tool. The objects are fed to the artificial neural networ k with two neurons in a hidden layer, and the result is compared to th e targets which are in our case equal to the inputs themselves. After training one can plot the objects' labels to the map whose coordinates are the output values of the two neurons of the hidden layer. The clu stering ability of such an arrangement is compared to those of Kohonen network feature maps and Principal Components Analysis using a well-k nown set of data, namely the set of 572 Italian olive oils. The result s of the presented method seems to be better than those of the PCA. Th e power of the new tool lies in the possibility of coping with nonline arities and in its continuous nature.