APPLICATION OF ARTIFICIAL NEURAL NETWORKS TO CHEMOSTRATIGRAPHY

Artificial neural networks, a branch of artificial intelligence, are c omputer systems formed by a number of simple, highly interconnected pr ocessing units that have the ability to learn a set of target vectors from a set of associated input signals. Neural networks learn by self- adjusting a set of parameters, using some pertinent algorithm to minim ize the error between the desired output and network output. We explor e the potential of this approach in solving a problem involving classi fication of geochemical data. The data, taken from the literature, are derived from four late Quaternary zones of volcanic ash of basaltic a nd rhyolithic origin from the Norwegian Sea. These ash layers span the oxygen isotope zones 1, 5, 7, and 11, respectively (last 420,000 year s). The data consist of nine geochemical variables (oxides) determined in each of 183 samples. We employed a three-layer back propagation ne ural network to assess its efficiency to optimally differentiate sampl es from the four ash zones on the basis of their geochemical compositi on. For comparison, three statistical pattern recognition techniques, linear discriminant analysis, the k-nearest neighbor (k-NN) technique, and SIMCA (soft independent modeling of class analogy), were applied to the same data. All of these showed considerably higher error rates than the artificial neural network, indicating that the back propagati on network was indeed more powerful in correctly classifying the ash p articles to the appropriate zone on the basis of their geochemical com position.