Utilization of Chemical Structure Information for Analysis of Spectra Composites
Kristin Domaschke, André Roßberg, Thomas Villmann · 2014
Abstract. In this paper, we propose the utilization of structural infor-mation of spectral data during the preprocessing to extend the ability of subsequent analysis methods. Specifically, we expect a dataset of measured spectra containing mixtures of only a few spectral components. Using the concentration ratios for a small subset of mixtures and the chemical struc-tural knowledge, theoretical spectral components are generated. Then a set, which combines measured and theoretical spectra, is analyzed using a self-organizing map to predict the unknown mixture ratios of the remain-ing subset by an associative learning. At this time, the initial study on simulated data reached very good results. 1